Use CaseInsensitiveFuzzySearch for repos with less than 80k files

Created by: olafurpg

Previously, WordSensitiveFuzzySearch was used by default for all repos. The benefit of WordSensitiveFuzzySearch is that it supports low-latency fuzzy filtering even for large repos with >100k source files. However, WordSensitiveFuzzySearch is probably more strict with capitalization than what most people expect from a fuzzy finder.

This commit adds a new fuzzy finder called CaseInsensitiveFuzzySearch that is enabled by default for all repos with less than 80k source files. This fuzzy finder is more relaxed with capitalization compared to WordSensitiveFuzzySearch. However, it has worse performance in very large repositories (example chromium/chromium) so it's only enabled for repos with fewer than 80k source files.

Demo: relaxed capitalization. The new case-insensitive search doesn't require you to capitalize the query or insert / characters to split up file parts.

Demo: refine the search with space characters. The new case-insensitive returns results that score best against all parts of the space-separated query. The word-sensitive filter would have returned 0 results after the first space character.

Big thanks to @mrnugget for suggesting the idea to use space characters to refine the fuzzy search. I really like this behavior!