Non-viable characters are a must in a good RPG and often, with enough intelligence and creativity these non-viable builds are viable and very fun to play. Having more options is always a plus, the fact that only 1% of gamers will appreciate it is another matter.
It's not a plus if it costs development time that could have been used better elsewhere. Making games for the one percent who want to RP crippled Dwarf Wizards is great in an ideal world, but in reality, if you want to keep making games and not fold (as they did) you have to prioritise--at least aim for features for the 5% and not the 1%. And you might say "it's just an option, it takes less time to allow it than to code it out" but the more options you have the more time you need to spend in QA etc., or else ship with piles of bugs.
It's not an accident that the studios who do this kind of stuff either go under, or else learn to stop doing it. Those are the choices. I loved KotOR2 way more than I liked PoE, for example, but if the choice is nu-Obsidian or nothing, I prefer nu-Obsidian. Something OK, but not great, is better than nothing at all, especially since at least there's still potential there to do something good in future. I love games by MCA, but if I were running a game company and it was my own money and my own economic future at stake, and the choice was hiring either MCA or Sawyer, I'd choose Sawyer every time. (At least until MCA shows he can take charge of a project and ship it successfully, in reasonable time and in good condition, something he's never done and is now avoiding having to do).