Hagen, that's interesting. I'd be very interested to hear others' views. It seems convention that misprinted clues don't need to make much sense after correction (albeit this gives the lie to any notion that they were accidentally misprinted which I thought was the conceit back in the early days - I could be wrong). It is of course that much smoother if they can make sense after correction, which is easier when they don't have to appear in every clue. And of course, it's even harder to do that where there are additional clue constraints.
I guess the setter's focus is mainly on disguising the misprints as well as possible, which adds to the difficulty further. I'm just looking at a recent draft puzzle I've done with misprints and there's no denying that the clues I like best are the ones where there is a genuinely strong surface reading after correction.