I'm not sure this ship meshes with your stated tactics.

For starters, it is anything but lightly armoured - about 25% of its wet mass is armour, that's 1.5-2.0x stock Gunship's armour mass fraction and Gunship is already a ship strongly oriented towards close-in gun battle.

My own similar vessels have about 0.5x Gunship's armour mas fraction, despite being built for direct combat.

In any case IMO it would be a good idea to optimize at least some guns for defeating missiles and drones - you can finish a crippled ship with primarily anti-missile CIWS just as well.

However, I think you are right that aluminium radiator is too easy to melt, I'll create a new light-weight radiator for that.

The tactic/strat/doctrine I used when playing CDE is avoid gun battle at all cost, intercept missiles/drones mid-flight, use missiles/drones to destroy enemy, escape if mission failed, use flares and guns as last resort. In fact, with common sense plus this doctrine, guns aren't really that necessary. It would be more effective to be construct a laser-only fleet guarding few and large but highly maneuverable high delta-v missile/drone carrier.
But guns were added just in case of:
a). Missiles are precious, using guns to finish enemies off can conserve ammo reserve. Remember in WW2 subs often use deck guns to finish off convoys after escorts were destroyed?
b). If entire enemy fleet decided to force a gun fight by burning straight at us. Then the lightly-armored missile/drone carrier fleet will stand little chance against them, so guns and armors were needed after all.

1. I agree with the mini NTR might be impractical. Would love to build some alternative to that.
2. For this point, I would argue modern guided destroyers even aircraft carrier alone are pretty much helpless against swarm of missiles and piles of conventional guns (assume the ship is in range of them). But the over all effectiveness of a weapon must be evaluated base on the strategies and tactics it is designed to employ with. If one uses the modern weapon with World War 1 doctrines (recall Iraq-Iran war), then of course it never satisfies the need because they are used beyond spec.

Ok, two points:

1. Main reason why your Candles are quite scary is that they are effectively KKVs on top of being nukes. They can expend their (considerable) dv budget quickly, at which point, even when shot down they are capable of taking out a warship simply by impacting it. They are also manoeuvrable and seem to be able to follow through hull breaches created by others and explode inside. Good thing there are decoys and counter-intercepts. I'm still unsure if NTRs on misiles would be practical IRL, mind you.

2. Your ship seems pretty helpless against missiles and conventional guns. It doesn't have much in the way of PD, aluminium radiators are also easily flashed off by nukes and streams of heavy bullets can collectively cut through the armour.

I honestly do not know why is there a seoncd Gunship 2...

About the starving part, I would like to have some power reserves in the first place.
Secondly, I had tried utilizing maximum power, but power hungry guns simply takes all the power to charge for themselves before less power hungry guns were able to fire. (Especially when using lasers, I wanted to have 3x130MW lasers, but they eneded up starving all other guns and made them not firing at all)
This created a awkward situation where power hungry guns spending long time to charge while high ROF smaller guns had to go silence and wait.
This ultimately reduces fire power outputs and I decided it is not worth it.
If I ever need extra guns (or missiles), I prefer to add another ship instead.

In this case the idea is more or less the same.
If they want build spaceships, they won't build a single 100kt behemoth.
One lucky shot will be enough to lose the battle.
Also the armor and guns are for final phase of combat.
When delta V is really low or somehow the fleet is caught in gun range, then the armor will become revelant. Before that, the fleet should use missiles or drones (if available) as much as possible. (Or when all enemy ships are disabled and need to use guns to finish them off)

Imagine USS Ford going gun battle with PLAN destroyer Type 54, no missiles.
No matter who win, final result is not gonna look good and people back home will complain: why not use planes/missiles!

I agree, but given the complexity of the ship already, I would say this is good enough and any other capabilities should be delegated to other ships.
In fact, in the unmanned derivative of this gunship I actually got rid of lasers entirely (while reducing mass down to 9kt); there will be a separated, dedicated and cheaper laser ship variant to provide laser fire power (planned to carry 2x130MW laser, 80cm aperture, 10 times more powerful than stock; with 8 high accuracy guns for defence).
My thought used modern Aircraft Carrier combat group as reference. Militaries don't stack all the missiles on aircraft carrier alone. They produce bunch of cheapter destroyers and stack hundreds of missiles on them.

My own strategy with crew radiators is to use small, refractory material ones and place as many as possible without bloating the maintenance requirements. Like you said, they're tiny screws compared to the rest of dry mass, so sprinkling them liberally is not that much of a mass penalty.

Personally, I avoid placing CM's far from center of gravity - this allows hard manoeuvring and surviving all kinds of things that can knock ship around - nukes, flak impacts, ammo explosions, propellant tanks rupturing, etc. I do place reactors in more than one location, though, as they handle high G better.

A point of note:
Even your armour can be burned through if someone really wants to get at the radiator on the opposite side. And that's with stock lasers.
Anyway, my own philosophy in regards to weapons is to have enough of given type to be able to saturate powerplant's output (barring other considerations, such as mass). You can always switch depending on the circumstances, if you don't want them starving each other, OTOH you don't want to be in a situation where you think "damn, if only I had one more laser" and you do have to lug the powerplant and its radiators anyway, so it makes sense to reuse it for as many weapons as possible.

I like the consideration you have given to mass distribution and tank draining order.

Woah, that's a lot for I wanted to talk about.

Also, in realistic scenario, I imagined a strategy for operating this ship is to drain rear fuel tanks first and leave front-most armored fuel tank as full possible for contingency; forward shifting center of mass will likely improve turn around time as well (This will be used in my short story to distinguish experienced ship commander vs unexperienced fresh).

So, yes, I agree with the flaws you mentioned. But the ship was not intended to be invincible, like modern guided missile destroyer in gun range combat with WW2 destroyer, it will work, just not good enough.

These are my design ideas explained.
As for the laser, I don't have a good design for it yet. And having 3 of them will starve other turrets, too.

The last phase, phase 3 is the situation where the gunship will be forced through 3 stages of combat and it will be critically damaged beyond combat capable.
The goal of this final phase was to ensure the ship still has capability to maneuver around or break orbit to receive help. 2 out of 3 crew modules were expected to be lost in combat, along with all propellants after the center line.
Weapons mount were placed to minimize projectil hit to the front-most crew module and armored fuel tank next to it; if enemy decided to disable weapons first. So that, when combat is over, at least 1 crew module and fuel tank will remain intact. Hence the need of a sturdy armor is necessary to ensure crew's survival.

Hence the choice of having 4 aluminium radiators. Also being cheap and easily replacable when damaged, since 90 kg radiators are tiny screws comparing to a capital ship and I would expect in realistic scenario, they will carry a lot of backups of these; this will be important in phase 3 of design ideas.
With all that said, I would expect gunship to break away as soon as possible to avoid transition into stage 2 and 3. In that case it is guaranteed lost of combat capability once battle is over, whether victory or not.

3. Death Zone, the final phase of gun range combat I had considered was the range where all weapons can hit the target, laser is producing formidable damage and orientation no longer provides adequate protection against projectiles. Like you mentioned, crew module's radiator is indeed an Achilles hell. I had considered how to mitigate that, but given the standard combat setup I use: 2 Defence Frigate, 2 Assasult Frigate, 1 Laser Frigate 2(based off gunship), 1 Gunship 2. If multiple enemies still alive at phse 3, projecttiles can come from many directions, having multiple radiators simply won't help in this case; if only one or two enemies survived. Then projectiles can come from relatively predictable direction then the ship can orient itself such that only one side of radiators are exposed (CDE's AI don't do this, just like when AI using sentinel). Two radiators on the opposite side should at least kept crew module partially functional.

This is also the phase where missiles should be fired, since ship orientation is relatively stable and minimizes the risk of missile crashing into armor belt. Missile launchers were placed behind a turret so it can survive longer (it was intended to be placed inside the dent after it, but I chose to use that to protect radiators, the missile launchers were not expected to survive in this case anyway).

2. Projectile weapons are now able to catch up the maneuver of gunship, since the transition between 1 and 2 varies depend on enemy's weapon, so armor have to take a ton of beats in case commander made mistakes. Once transition had happened, the ship should orient itself head-on, ideally into the stream of bullets and utilize the armor to withstand enemy fire. This phase also applies if the ship encounters flak missile, it was designed to orient itself toward flak missiles. Certified to withinstand 60 small flak missiles (20kg) approaching at 5km/s with only a scratch, although the ship will be tossed around very hard due to the impact force.

Second phase is gun range combat, where either enemy or friendly is now within opposition's gun/laser firing range; I only considered low velocity intercept since high velocity intercept sounds good on paper but not practically useful in most cases (@SR-71, it was great for its time, except it isn't). This low velocity sand blasting phase is further divided into 3 stages.
1. When combat started, not all weapons are in range and most projectile weapons require time to travel and laser intensity drops in squares relation per unit of increased distance. Threat exists but is minimal at this point.
So, my idea was to use agility and acceleration to dodge incoming projectils and even out laser damage.

The design idea itself included three phase of combat.
First one is long range combat where it utilizes missile and mobility to stay out of range of enemy while accompanied by Defence Frigate (sold separately) to defence itself against long range threats; to minimizes chance that the ships get caught in gun range, and to maximize mission success rate, Candle was designed to be overkill from the start, better to waste ammo than lose life or a capital ship. It is also intended to be used to kill laser ships via its small profile and shear numbers.

I agree with the radiator part. However I would argue even with smaller radiators, if it ever get caught in gun range, low profile/small radiators will not last long either. Hence the reason of 18 radiators to provide extra redundancy and placing radiators for crew module inside dents to minimize chance of being hit (when head on).
Tp explain the armor scheme and radiator choice and placement, I had to explain my design philosophy.
I was intended to write a short story involving space combat and I wanted a realistically designed battleship and I had a design idea in mind.

Also, it now has more than enough juice to power more than one 100MW laser - why remove the other one? I would throw in the third one instead.

The problem with dependencies is that if anyone uploads a module with the same name, they will be happily overriding each other and breaking stuff.

Nice ship, as an incremental upgrade to stock gunship. I like the look of low-profile radiators too.
That said, aluminium crew module radiators are a huge Achilles heel, so is keeping all the reactors in the (open) tail end.
Candle missiles are quite formidable, although I consider NTRs on missiles (or non-subcap drones) a bit of an overkill. The launchers could use redundant radiators as well, especially given that other weapons aren't as scary.
The armour scheme is nice, but seems overkill - it will hold long after the ship is killed by loss of external modules, slugs flying through turret and launcher holes or even spallation.

Also, after some tweaks I can now kill this ship without much risk with my stock modules "Dragon" - by focusing on taking out launchers and crew radiators first.

I'm not sure how dependency items work, can people use this ship without errors? Or I must set dependencies myself?