The way I see it is: it's not the judge's job to decide whether something is solvable or not, only whether the puzzle you have in front of you is solved. (I am not a delegate, have never organised a competition, have never judged, and am not well versed in the history of the regulations. This is my take, which may or may not be the actual reason for why the regs are the way they are.)
With a puzzle state that's one move off from being solved, it's (usually) clear to anyone that it's one move off. (Exception: square-1.) It happens to be true that if you have a corner twist, that's unsolvable, but judges need not be cubers themselves and even if they are, they need not be aware of whether such states are unsolvable.
What if, during a 444 solve, you suddenly forget how to do OLL parity and you finish with a flipped edge pair? A judge unfamiliar with big cubes might mistakenly assume that it's unsolvable, but that's not a decision they should be concerned with anyway—they should only be looking at whether the puzzle is solved (more precisely, whether it's solvable in one move), and it's evidently not. That's a much easier decision to make.
(Most cubers don't even know why the unsolvable puzzle states (e.g. single corner twist / single edge flip on 333) are unsolvable anyway, at best having heard a partial explanation from someone else. How would you convince someone that a corner twist is unsolvable, versus being just a legal position that you haven't learnt how to solve? There are puzzles where a twist of a single piece can be solvable (e.g. pyraminx).)
www.worldcubeassociation.org
