ansi2json: Handle multiple CRs in line endings

On Windows-based CI pipelines we can get text where line endings consist of multiple CRs followed by a LF. That's because many programs use LF line endings internally and convert to CRLF on output. That conversion is done without any check, so if the string is already CRLF we get CRCRLF. That's the case of the Microsoft C/C++ Runtime Library, CPython, etc.

Handle that in ansi2json by matching multiple CRs followed by a LF.

before	after

With this MR the Python's traceback is correctly shown

See https://gitlab.gnome.org/GNOME/glib/-/issues/3464 for more informations

The issue can be reproduced by executing this Python script in a Windows job:

import subprocess

result = subprocess.run(['tar', '-tf', 'filename.tar'], capture_output=True)
print('--- stdout ---')
print(result.stdout.decode('utf-8'))
print('--- stderr ---')
print(result.stderr.decode('utf-8'))

Helps #218771, #342686 (closed)