Simplify document class list detection

[thecrypticace]

This should be mostly functionally equivalent. It simplifies a good chunk of the code used throughout to find class lists and class names.

1. It eliminates many repeated calls to getConfiguration(…) which would increase based on the number of embedded languages in the document. These hit a cache anyway so the overhead wasn't significant but the repeated calls aren't necessary.
2. Because of this I've now made several things synchronous where they were async before.
3. It eliminates a search range limitation for hovers which would cause some class name hovers to fail on large enough documents depending on document structure and what came after the class name further in the document.
4. This will simplify a future refactor that caches language boundaries, class lists, and class names on a per-document basis and only recompute when documents change.