Fix spurious "Output was trimmed" message in archive.extracted

[ggiesen]

What does this PR do?

archive.extracted appended . Output was trimmed to {} number of lines to the state comment every time files were extracted, regardless of whether any trimming actually happened. Trimming only occurs when the trim_output argument is set, so with its default value (trim_output=False) the message was still added with the literal value interpolated, producing the misleading comment Output was trimmed to False number of lines.

This wraps the comment append in if trim_output: -- the same condition that already gates the actual output trimming -- so the message is only shown when output was really trimmed. By that point trim_output has been normalized (True -> 100, string integers -> int), so the value shown is always the accurate line count.

What issues does this PR fix or reference?

Fixes #59570

Previous Behavior

Running archive.extracted on a source that actually extracts files, without specifying trim_output, produced:

Result: True
Comment: <archive> extracted to <dest>. Output was trimmed to False number of lines

even though no output had been trimmed.

New Behavior

With the default trim_output, the trailing "Output was trimmed to ..." sentence is no longer added:

Result: True
Comment: <archive> extracted to <dest>

When trim_output is set to a truthy value and output is actually trimmed, the message still appears with the correct line count (e.g. Output was trimmed to 1 number of lines), unchanged from before.

Note: the issue discussion also raised a larger redesign of trim_output (accepting only integers and treating "unset" as "don't trim", since it currently accepts both bool and int with True meaning 100). That is a breaking API change that would need a deprecation cycle and is intentionally left out of this fix; this PR resolves the reported symptom without changing the argument's contract.

Merge requirements satisfied?

• Docs -- n/a (behaviour-only fix, no documented behaviour changes)
• Changelog -- changelog/59570.fixed.md added
• Tests written/updated -- added test_tar_bsdtar_without_trim_output_59570 (default trim_output, asserts the message is absent) and test_tar_bsdtar_with_trim_output_zero (explicit falsy trim_output=0); the existing test_tar_bsdtar_with_trim_output (trim_output=1) is retained as the must-not-regress case proving the message still appears when trimming really happens.

Commits signed with GPG?

No