[bpmore]

This fixes defects 1 and 2 reported in pestphp/pest#1942. It is the fix offered in that issue, unchanged, with tests.

What changes

• src/Support/NodeVisitor.php: the captured tail is trimmed before the empty test. Previously // @pest-mutate-ignore (with a trailing space) and /** @pest-mutate-ignore */ produced a non-empty tail. That tail matched no mutator name, so the annotation suppressed nothing.
• src/Support/MutationGenerator.php: $mutatorsToIgnore is reset for each annotated line. Previously, once any earlier line used the @pest-mutate-ignore: form, a later bare annotation was recorded with that earlier line's mutator list, because ?? ['all'] could no longer fire.

The meaning of a named list is unchanged: an unknown mutator name still suppresses nothing.

Tests

Added to tests/Unit/MutationGeneratorTest.php:

• a bare annotation with trailing whitespace suppresses every mutator on the statement below it;
• a bare single-line doc block suppresses every mutator on the function;
• a bare annotation after a @pest-mutate-ignore: IfNegated line suppresses every mutator, rather than only IfNegated;
• an unknown mutator name still suppresses nothing, in three forms: plain, with trailing whitespace, and in a doc block.

With the two source changes reverted, the first three fail and the unknown-name cases still pass. Reverting NodeVisitor.php alone fails the first two. Reverting MutationGenerator.php alone fails the third.

The full suite on PHP 8.4, macOS, reads 3 failed, 322 passed, against 3 failed, 316 passed on 5.x without this change. The 3 failures are the same visual snapshot tests on both. composer test:types reports the same 2 errors before and after, in AbstractFunctionCallUnwrapMutator.php and RemoveEarlyReturn.php, which this change does not touch. Pint passes.

Not in this pull request

• Defect 3 in [Bug]: [mutate] @pest-mutate-ignore silently suppresses nothing, or the wrong set pest#1942 (an unchecked file_put_contents into a .temp/mutations directory that is never recreated) is a separate mechanism and is not fixed here. That issue therefore stays open after this merges.
• // @pest-mutate-ignore: Name because reason stays inert, exactly as before. Whether to accept that form is the open question in [Bug]: [mutate] @pest-mutate-ignore silently suppresses nothing, or the wrong set pest#1942 and is not decided here.
• Windows line endings in the line scan ([Bug]: [mutate] Windows: @pest-mutate-ignore comments land on line 1 in files with LF line endings pest#1928, fix: support LF line endings for @pest-mutate-ignore on Windows #44) are a different defect and are untouched.