fix #30394, an unsoundness in ml_matches

[JeffBezanson]

This fixes a corner case where a bug is caused, counter-intuitively, by an over-estimated intersection.

Here's the setup:
We have method signatures A and B, with A<B (A is a strict subtype). We have a dispatch tuple X, where X<:B and !(X<:A). However, intersection returns X for intersect(X,A).

Since there appears to be a match with A and A<B, ml_matches skips the match with B. The fix just requires dispatch tuples to be a subtype of a signature in order to match at all. The unsoundness itself is actually in the logic just below the new code which sets skip = 1.

[JeffBezanson]

I'll note that the logic in question:

                    jl_subtype(closure->match.ti, prior_ti)) {
                skip = 1;

can still be unsound: prior_ti is a type intersection result, and the bigger it is the more likely this is to be true and so the more likely we are to incorrectly skip a match. However I think other variations of this bug are a bit less likely to occur, and also seem more expensive to prevent, so I'll leave it for future work.