Almost all of the tie-in fiction I've ever read was incredibly mediocre from a writing standpoint: everything from the Dragonlance novels, to R.A. Salvatore's Forgotten Realms work, to David Gaider's Dragon Age novels. There may be a fair number of reasons for it (the authors in question were still learning their craft, strict deadlines and the necessity of being involved in multiple projects at once interfered with the process), but I think the main reason may be that a lot of tie-in authors are gamers first and authors second.
I haven't read the Dragon Age novels, but from what I've seen of Dragonlance and R.A. Salvatore's work, I am in absolute agreement with your assessment of their mediocrity. Tie-in fiction, regardless of the circumstances surrounding its creation, carries a negative stigma for good reason - it's telling how uninspired the first few books of the Dragonlance series (all that I bothered reading, and then only cause they were in an omnibus edition and I figured I might as well finish it) felt overall, apart from a few brief moments where it seemed to transcend its roots in AD&D, when Weiss & Hickman's Death Gate Cycle set in their own original IP is brilliant, and remains to date my favorite fantasy series of all time ever since I first encountered it.
So what is it about tie-in fiction that stifles the creative output of good authors? If I had to wager a guess I'd suggest the problem stems from trying to echo conventions or elements of the original property to the point where it becomes detrimental, and a question of motivation and "purpose".
Take the Dragonlance series for instance - a large part of the problem with that series can be directly linked to the conventions of the source material getting incorporated directly into the narrative structure of the setting. The rules of D&D, like those of pretty much all roleplaying systems, represent abstractions - wizards can cast x amount of spells of y level per day for game balance purposes more than anything else. We don't question that when it's an element of a game we're sitting down and playing whilst rolling dice and cracking wise, but really, the "you can cast 4 spells today, no more, no less" mechanic is silly - spellcasting wouldn't really work like that in the actual universe we're pretending to be in while we play the game, it just works like that because it's a game.
Except in Dragonlance that's exactly how spellcasting works, and it's every bit as jarring and baffling as it would be if, during the battle for Helm's Deep in the film adaptation of The Two Towers, the participants suddenly started taking turns rolling a d20 to see if they hit each other. There's a reason characters in video games don't generally start openly discussing hit points or which button you need to press to do various in-game actions, and that's because having characters acknowledge game mechanics only calls attention to the fact that you're playing a game, and that spoils immersion.
So when authors faithfully recreates things that we only don't call into question from the original game now because it's a game in their novel(s), it rings distinctly false; some things simply should never survive the translation across mediums unaltered. Nobody in their right mind would sit down to watch your average FPS game and then novelize exactly what it was like to watch somebody playing an FPS game, because that would make for a shitty novel - books aren't FPS games, having your protagonist always wind up by himself due to to contrived circumstances, witness "cutscenes" through glass barriers, receive all his exposition through a headset mic, collect "tri-part keys", and other conventions of FPS games, in a book, is just plain old bloody stupid.
If my complaints above seem a tad... specific, well that's because such a book exists, and it was awful in a way that no tie-in fiction I had ever read before could boast: Fire Warrior is essentially a guide to all the things you should never do as an author of tie-in fiction for a video game, made all the more disappointing because the fellow who penned it is actually a decent author. Everyone looking to get into that field really needs to read that book and take a lesson from it.
Possibly surprising revelation given what I've typed so far: Almost 100% of the fiction I read these days is comprised of tie-in novels, and I'd contend that most of it is either fairly good or even excellent, because what I read is Warhammer 40,000 novels. See, a lot of the problems endemic to tie-in fiction don't really plague the world of 40K novels - authors don't attempt to foolishly replicate the rules of the table-top wargame into their fiction for the same reason that the official background of the setting doesn't, namely because that would be stupid; 40K's overarching "plot" is "it's the future, and there is only war", and the setting is the entire bloody universe - authors have a tremendous amount of leeway to write about whatever they want, to destroy entire star systems, to invent them, you name it.
The 40K novels end up feeling more like the products of a shared universe with common thematic elements and rules that various authors more or less adhere to than it does your typical "tie-in fiction", precisely because in almost all instances the novels themselves don't tie into anything beyond the 40K setting/aesthetic itself; the few novellas and longer works that are direct product tie-ins for corresponding video games or GW products are invariably markedly worse than the "original" books released under the 40K umbrella.
Authors like Aaron Dembski-Bowden and Chris Wraight are producing some truly phenomenal material, can't recommend them highly enough.