Given that this is an RPG, there is plenty of reason to suggest that making decisions like a player would even with limited information should not be done by denizens of the world.
AI not being able to prioritize targets based on their defenses and threat level due to having no information and thus no ability to estimate such concepts while the player can do so even while lacking precise information creates a massive disparity in tactical adaptability of both parties which heavily favors the player and creates the need to strengthen enemy units in order to compensate for suboptimal behavior. There are only two solutions to this problem: provide precise information to both parties which is fair or provide limited information to the AI while leaving the player guessing based on visual or some other (e.g. narrative) ques. Your statement that no information should be provided to either party is not based on any argumentation, it's based simply on your preference and that preference leads to dogshit design - compensation of stupid behavior via stat bloat.
Mechanics can also be made by improving the behavior of line of sight to further limit player information
Which solves nothing, because players can make intelligent guesses even with limited information. No matter how hard you gimp the player in terms of provided information, people will still exhibit intelligent target prioritization based on available ques. Deny this to the AI and you get an artificial idiot acting in a predictable manner and thus removing the need for tactical adaptability, i.e. removing the tactical depth from the game.
information a player should not have
This is not supported by anything other than your subjective preference. The need to have information provided to both parties is supported by the fact that this evens out the playing field (as much as it's possible with AI) in terms of intelligent target prioritization between the player and the AI.
Closest targets are often the most logical targets
No, they aren't. The most logical targets are those that are the most convenient to attack based on their defenses and threat level.
Enemies will prioratize low AC characters in this game if they are close enough while initially attacking.
Wrong and easily proven via archer behavior. Archers will attack mostly closest targets even if they have lower AC targets in sight. Target prioritization may or may not occur if melee enemies (usually with very long reach) have multiple targets in range, but this is never guaranteed even on unfair with super long reach characters like Deskari. If these same melee enemies start attacking one high AC character and then you move a lower AC character into their range, no target prioritization occurs unless the enemy keeps missing the target for many turns and only in RTwP mode. I have not observed this behavior in TB mode.
I think the problem is that you haven't played these games since Kingmaker and are extrapolating based on Kingmaker.
The problem is that you have no clue what you're talking about and keep getting game behavior wrong, caused by your lack of game experience.
Most of them do not have a line of sight mechanics either.
"Most" means nothing when there are plenty that do. JA2 and UFO games are considered as some of the best in the genre an they have both line of sight and friendly fire mechanics. Even newer TBT titles like Mordheim have line of sight mechanics as well.
None of them have line of sight mechanics.
Wrong. Japanese SRPGs as old as Hoshigami have both line of sight and friendly fire when firing bows.
Not for any tactical reasons.
The tactical reason is to maximize hit chance and avoid friendly fire.
Not mostly scripted. Sometimes it occurs in random encounters.
Precisely mostly scripted, precisely mostly after a cutscene when they are scripted to target the MC. Or in cases when specific enemies are scripted to attack a particular character like the Sarkorian Kinslayer attacking Ulbrig. Archer target prioritization is based mostly on proximity, the cases when for some unexplainable reason they choose to ignore the first closest target is presented to them are exceedingly rare and do not follow any pattern.
line of sight is required and can be implemented with individualized fog of war
Individualized fog of war cannot be implemented for the player who controls the entire party and has the visual information from every character. If you chose to implement individualized fog of war for the AI while the player sees everything (and the player will see everything unless you have different people controlling different characters and even then players could communicate), you're just gimping the AI, creating another instance for the need to compensate for suboptimal AI behavior via stat bloat.
Except the "tactical depth" in this game is just playing the caster the way you should already being playing it. If the counterplay is just playing the same way as normal, that is buffing everything, it's not really counterplay.
Wrong. Using spells of level 2 and 3 on protections such as Mirror Image, Communal Protection From Arrows and Displacement, when those same spell levels can be used for devastating AoE disabling spells such as Winter's Grasp, FavoriteSelective + FavoriteExtended Sound Burst and Stinking Cloud + Corruptor, in cases when you don't actually need those protections is a waste and suboptimal play. You lack the insight how to play a proper caster.
But they do step on them.
Yes, because Owlclowns are incapable of coding proper pathfinding. As I said, that would be the perfect solution. The only problem with traps stems from developer incompetence, enemies not triggering traps is not a problem, because in the hands of a competent developer they'd have the pathfinding to walk around them.
They're not that rare if you exclude random encounters from consideration. Most areas have at least one or two places where there is a set of traps with enemies that engage you while you're in the proximity of the trap.
Wrong. They are exceptionally rare. "Proximity of the trap" means nothing when the trap is out of the player's way used to approach the enemy. In-combat instances where traps are placed specifically on the only path players can use to approach the enemy are exceedingly rare. I already gave you an example: Currantglen underground on the 2nd visit where you need to approach a group of scavengers near the altar has a trap placed in a chokepoint that is unavoidable by a party moving in formation and can only be avoided by a single character carefully maneuvering around it. These are the traps that matter in combat, simple "proximity" means absolutely nothing as long as you can find the trap.
There are also quite a few random ambush encounters possible in WotR where your party starts with enemies surrounding you or behind you and the enemies have a suprise round.
Correct, and in those cases target prioritization still plays by the same proximity rules already covered above. Since the party members start surrounded, you get enemies targeting multiple party members based on their proximity.
They are already used quite liberally.
Wrong, they aren't. Traps that inevitably need to be maneuvered around in combat due to being placed on critical approach routes or in chokepoints are exceedingly rare.
Have nothing to do with teleportation and are instantly spotted by any properly prebuffed party. Not even mind-blank will save them from echolocation.
Teleportation in most cases requires access to at least 4th level spells within the 3e/pf ruleset. Usually this requires a standard action to implement. BTW, WotR already has enemies that do this like some of those Babau. But then your argument is that Owlcat is still all prebuffing?
Access to teleportation is a question of enemy character builds, including access to scrolls (something Kuroisan the Acid Kensai tormented players with in BG2 Tactics two decades ago and still cannot be properly copied by professional devs) or wands via UMD. Casting dimension door as a spell via swift action is a question of access to quicken rods or metamagics. You can solve the "problem" of creating such enemies in various ways, including just letting them cast it as a standard action and not having them attack on the same round for a weaker version of an enemy archetype. Yes, there are babau Infiltrators that I have already mentioned, which are exceedingly rare and are only present in a couple of places in the game. I have nothing against prebuffing enemies with actual in-game legal buffs, but that's not what Owlclowns do: they bloat enemy stats relative to tabletop in a manner that cannot be countered via dispel and they do this specifically to compensate for their inability to design encounters in a smart manner.
Designing encounters just around killing backliners is also bad design. Why? Because all you have to do is leave the low AC characters at the entrance of the map while the high AC characters who also have high saves, high AB, and high damage, clean up everything as usual. No tactical depth required to beat this design.
Forcing the player to adapt to various tactical scenarios is good design, because it creates tactical depth. The goal is to not kill the player backliners, the goal is to force the player to adapt by switching up their tactics. If you leave your low AC characters behind, you have to fight with less characters which is a forced adaptation. Alternatively, you can just protect your low AC characters with multiple layers of defenses: mirror image + displacement or blink (in cases where enemies can negate or reroll your displacement). You have no clue how to set up defenses.
It just ends up making the caster characters and so on who are lower AC just not be used for anything but buffing
Absolute nonsense. Offensive casters are the strongest classes in any situation. If all else fails (which it shouldn't), a proper offensive caster can be set up to have unbeatable initiative and enough spell DC to disable entire encounters before they are threatened. As stated before, you lack insight on have to build casters.
Is your problem due to lack of game knowledge.
Shouldn't happen in that case.
That's exactly how it happens in WotR. If you advocate for implementing more advanced cover mechanics from TT, then I have nothing against it.
It's like you ignored all the Votaries and quite a few of the alchemist and caster normal enemies that start their fight by getting a dozen quickcasted buffs.
Like I said before: the vast majority of enemies in either Kingmaker or WotR are unbuffed (via legal in-game buffs) trash mobs. Of those exceptional few regular enemies that are buffed, most are buffed with only a few spells even in the endgame. The proper prebuffing treatment is granted almost exclusively to bosses and mini-bosses like the Desolating Gallu Stormcallers. Instead of bloating enemy stats in a way that cannot be dispelled, they should be buffed via legal in-game buffs. The resultant loss of encounter challenge should be compensated via improvements to AI and encounter design.