Frizbee does not match some subsequences in skim

[LoricAndre]

I use a file containing 1M "file-like" lines to benchmark and check skim's matchers. When running tests, I found that frizbee seems to miss some items that are subsequences when using max_typos: Some(0). I use a Default scoring except for the matching_case_bonus which is set at 0 here.

The query used here is test.

frizbee-mismatches.txt

[saghen]

Pushed a change that fixes the score matrix rows not getting zeroed out correctly, which increased the number of matches in my chromium test. There might still be some bugs I'm missing for this. I need to bring back my reference implementation :)

[LoricAndre]

Yeah, correctness is really hard to test for with these algorithms