gh-154470: Fix spurious ^J in pdb's colorized list command

[ambv]

pdb's list colorizes each source line with _colorize_code, which feeds it to _pyrepl.utils.disp_str. disp_str renders control characters in caret notation, so the trailing newline of each source line became a literal "^J". _print_lines only stripped the line after colorizing, and rstrip() cannot remove the embedded caret sequence.

Strip the line before colorizing, matching what where already does via format_stack_entry. Add a test that attaches with colorize enabled, runs list, and asserts no "^J" leaks into the source lines.

Broken by gh-146584.

• Issue: pdb -p: list command output contains trailing ^J markers #154470

[gaogaotiantian]

I agree that the better way is to strip before colorize. However the is the test more complicated than it needs to be? We don't need remote pdb to trigger this issue - a local one would do. We missed it probably we normalized the white spaces in doctests?

BTW I'm aware of this issue a while ago but did not get the time to fix it. Planned to fix before 3.15.0 official release. Sorry about that.

[ambv]

Yeah, the failure existed on local pdb, too, but I opted for a remote test, because coloring happens on the remote end, so this additionally tests that this survives the client-server connection. Is it a big deal to have a remote test here?

[gaogaotiantian]

I think it's nice to have additional coverage for remote pdb, but we should probably have a quick and clean test for local pdb - it would be much easier to write and it targets the colorization with more focus. The remote pdb tests could potentially fail due to many other reasons.