What makes more sense:
Incorporating an epub3 validator in an app that still primarily produces epub2 books?
Or continue to use an epub2 validator and make allowances for the handful of epub3 features that some retailers are cherry-picking?
I for one believe that it makes absolutely no sense to incorporate an epub3 validator into a primarily epub2 system. It's not like xhtml or css where using a stricter spec won't really hurt. You're talking apples and oranges. Most everything would fail validation miserably employing epub3 rules.