Turn-based combat is generally superior in games which rely upon heavy statistics and precise tactics, particularly wherein long-term planning is necessary over short-term reflexes and response. While it's possible to do this in a real-time situation, it is far more difficult, and almost always, the player's action skills take priority over strategic considerations. As a result, most action games tend to feature very simple combat by comparison - you can sometimes pull it off (see something like ArmA for shooters, Mount & Blade or Dark Souls for swordplay, etc.), but even then it rarely matches the depth possible when you aren't chained to your own twitch skills.
Additionally, unless you have great AI, chances are managing a party without a turn-based system is an extreme hassle; one reason I firmly believe later BioWare games are limited to smaller parties is precisely because the real-time nature of them would make any more than three or four very difficult to manage. Compare that to Icewind Dale, with parties of six custom-created characters, or Arcanum, with sometimes as many as eight, and you'll find that turn-based is basically the only way to go to ensure proper tactics and battlefield control. The word "battlefield" is key, by the way - the scale in a turn-based system (especially with an isometric perspective) is also generally going to be far greater than in real-time games.
It's also worth pointing out that it is way harder to simulate the nuances of combat in a real-time situation. How many of those games feature large battles with dozens of people or monsters involved simultaneously? What about different weapon types with their own distinct advantages and disadvantages, i.e. spears, halberds, staves, clubs and maces, long and short firearms, cavalry, cannons, and other stuff that tends to be "trimmed out" to save on budget and to make games more accessible? Sure, you can pull it off in a real-time context, but when you combine that with the need for modern-day graphical fidelity and accuracy as far as animation, physics, controls etc. goes, it becomes extremely expensive to try and build real-time combat around more than a single type of weapon. Imagine making a separate set of mechanics every bit as demanding as, say, the shooting in Gears of War, the melee combat in Ninja Gaiden, etc., and I think you'll realise that it just isn't very feasible without serious compromises that, frankly, most fans of real-time games would scoff at.
I also don't buy the argument that turn-based combat is less immersive. I get immersed into a game not so much due to great visuals and sound, but when I'm involved in a set of complex game mechanics which captivate and engage me. Good graphics and fluid controls and all that are important, but ultimately it's gameplay that keeps me coming back and gets me to say "just a few more minutes!", not how many hairs have been meticulously rendered on Commander Shepard's ass, or how much blood fountains up from a decapitated neck, or how much of a badass I'm made to feel like by a designer who thinks I need to compensate for something. Great turn-based combat is frequently far, far more engaging, exciting, demanding, and entertaining than hack-and-slash or shooty-shooty real-time combat can ever hope to be, because it's got the depth, nuance and challenge that so many of those other games lack.
That's not to say all real-time games are like that of course - I don't want to confuse the form itself with poor examples of it, and on that note, it's also worth mentioning that yes, shitty turn-based combat exists as well. Just keep in mind the distinction between the fundamental mechanics and the games themselves.
On a side note, I also just want to say that I almost always prefer grid-based turn-based combat over games that don't use grids. It can work, certainly, but not having that precise knowledge about whether a character is say, blocking a door, or not being 100% sure how far you can move in a turn always bugs me. That's why I tend to stay away from Infinity Engine combat - even though it's turn-based, it's still fairly imprecise and sketchy as to whether something will actually work or not, and it comes down to the interface and game design, not my skill as a player.
The fact is that all game mechanics are abstractions. We tend to hold up things that resemble real life as being more accurate and better, but I don't think this is a particularly healthy perspective to have. Games that simulate real life perfectly are very rarely fun (though this depends on the game and the individual), and we accept tons and tons of concessions in the name of entertainment. Why, then, hold one form of abstraction over another? It's like saying jazz is "better music" than classical - you can't directly compare them on any objective level and say one comes out on top. I can understand a person liking one and disliking another, but that's a matter of taste, not objective quality.