speedup has_comments_in_lines with binary search

[froydnj]

We don't hit this code very often, but we might as well try to be fast when we do.

This way performs a little bit more work than necessary -- there is probably a smarter way to do this based on binary searching for the start and then the end of the range, and then deciding based on those results, but this way is more elegant.

[froydnj]

Realized this morning that this is not quite right -- contains doesn't partition the slice in the way partition_point requires. The partitioning is FTF instead of FT. So we'll have to do the real binary search thing, which I'll write up later today.

[fables-tales]

I have a concern that we're moving from a data structure that is inherently correct for this problem to one where we have to be careful about how we use it. Are we sure it's actually worth it here?