In principle I feel like edits which improve a question, however minor, are still improvements and should be made by anyone who has the inclination to do so. This makes particular sense if your view of the site is not just the (increasingly) rapidly changing front page in which the latest questions are being answered, but as a repository of useful questions and answers for posterity. I think the latter is clearly more useful in the long run but harder to achieve: just look how often the same questions get asked and answered, for instance, before someone can track down the duplication.
I view the fact that even minor edits cause the question to get bumped as a flaw (or, if you like, suboptimality) in the platform. It is one of several site mechanics that seems more concerned with preempting some kind of pathological gamesmanship that has rarely or never actually occurred on math.SE. (Anyway, it would be easy enough to have minor edits be listed for review in a separate category and for people to flag moderators if they are done abusively.) At the moment the front page is pretty flooded as it is: I think it has reached the point where most serious users will miss questions of interest to them if they don't use the ignore and search options rather extensively (and I think that's a problem). Maybe bumps coming from minor edits will make things even worse; maybe things are already so bad that it doesn't really matter.
Always taking the philosophy to work within the site as it currently exists, I think the best current solution is for people to be very proactive about cleaning up questions in the hour or so after they are asked. A lot of people do this already (I noticed Michael Hardy doing it this morning, for instance). I think these people are doing us a great service and that even more of this would be even better.