fix reports the wrong number of removed ignores

[asukaminato0721]

fix #1309

[yangdanny97]

Hi, do you mind writing up a quick summary of the nature of the bug and how you fixed it? It would make the logic a bit easier to follow. I can see that it fixes the issue, but it would be useful to know why certain things happen.

[asukaminato0721]

old: incremented the counter for every candidate line flagged by find_unused_ignores, even if the line no longer contained a suppression marker.

new: split the work into' remove_unused_ignores_internal', returning a SmallMap<PathBuf, usize> of actual removals. This allows the public to log the correct totals and provides tests with direct access to the counts.

In the inner loop, now check the regex.is_match(line) before counting and mutating the line; untouched lines at the target positions are copied through without bumping the tally, so empty or whitespace-only rows don’t inflate the numbers.

[meta-codesync]

@rchen152 has imported this pull request. If you are a Meta employee, you can view this in D84954481.

[yangdanny97]

Review automatically exported from Phabricator review in Meta.

[meta-codesync]

@rchen152 merged this pull request in 1676458.