Rank() sorts non-matching items above real matches (e.g. "hell" ranks above "say hello …" for "hello") #123

What's wrong

TextFilter.Rank (TextFilter/TextFilter.cs:204-217) calculates isMatch for every item but never uses it. It sorts on score alone:

return items.Select(item => { bool isMatch = IsMatch(..., out int score, TextFilterType.Fuzzy); return (item, isMatch, score); })
	.OrderByDescending(t => t.score)
	.Select(t => t.item);

When the pattern isn't fully present, FuzzySearch's score only reflects how far the partial match got. A short near-miss avoids the penalties for skipped and extra characters that a real match inside a longer string pays. As a result, non-matches regularly score higher than genuine matches.

Keeping non-matches in the output is intended; the existing test keeps "world" for "helo". The bug is that they sort above real matches.

Reproduction

Observed scores from a scratch test:

Why it matters

Rank is the API for type-ahead and file-picker lists. Typing config puts conf ahead of the config.json you are looking for, and readme puts read ahead of readme.md.

Suggested fix

.OrderByDescending(t => t.isMatch)
.ThenByDescending(t => t.score)

The existing Rank tests still pass with this change, because "world" is a non-match and already sorts last.

Add a regression test: Rank(["hell", "say hello to everyone"], "hello") returns "say hello to everyone" first.