"Quasimodal" is a useful category since it denotes similarities and disparities. Grammaticalization as a concept can at least open the possibility that these constructions are on a path toward modal. We can also open up the possibility that grammar patterns exhibit some of the eccentricity (delicacy) we find routine in the lexicon.
One key, I think, with "used to" is not just that it doesn't have a present tense alternative for its modal like meanings, but in a strict sense does not have a nonfinite form (in the same way that "have to"and "be able to" have.) "Use to" isn't fused and has a different meaning. The exception, I guess, might be in the negation examples we have been discussing. The argument for "didn't used to" could include the idea that "used to" has modalized sufficiently to have lost its nonfinite form. That usage shifts back and forth might mean it is still in process and has different status for different people.
It's an interesting problem. I like the idea that most of us seem opposed to legislating this one way or the other.
Craig
Visit ATEG's web site at http://ateg.org/