fix: reset the snapshot counter between flaky retries

[lazerg]

What:

• Bug Fix
• New Feature

Description:

SnapshotRepository::$expectationsCounter is static and keyed by test file plus description, so several toMatchSnapshot() calls in one test resolve to .snap, __2.snap, and so on. Testable::__runTest() replays the closure in the same process on a flaky retry and never clears that key, so attempt 2 looks for __2.snap. It does not exist, gets created from the current value, and the test ends as incomplete on a green suite. Every run after that passes, because attempt 1 still fails against the real baseline and attempt 2 matches the file the first run wrote. On CI the run fails, but it names a __3.snap nobody created instead of the value that actually drifted.

The retry already restores object properties, mock objects and the output buffer before running again. The counter belongs to that cleanup, so it is reset there too, and the retried attempt resolves the same snapshot file as the first one.

Related:

Fixes #1871

[MrPunyapal]

Thanks! Already covered by b8c2265