match scorer methods closely

[philschatz]

The .match() method now returns an array of string indexes for each matched char. This allows the code to follow scorer almost exactly and simplifies the logic in fuzzy-finder.

It does have one issue. Since scorer will double-count characters, a search for rc against src/scorer.coffee will result in s[r][c]/sco[r]er.[c]offee. See screenshot:

This "feature" can be removed by simplifying the unless queryHasSlashes code in fuzzaldrin.coffee to either:

• return the baseMatches if all the characters match, or
• return the full path matches if baseMatches is empty

The corresponding https://github.com/atom/fuzzy-finder/pull/39 has been updated to use this code.

[kevinsawicki]

Thanks for this

[kevinsawicki]

I published 2.0.0 of this library with this in it.