Conformance scripts - ISM metadata file comparison seems to be line-ending-sensitive

Both the BE and the non-BE conformance scripts use filecmp.cmp python function for comparing files for bitexactness. ISM metadata files are .csv files - so basically textfiles - and will have different line endings between windows and Unix-derivates. At least in my test run with the BASOP BE confromance (see issue in BASOP: sa4/audio/ivas-basop#2234 (comment 99231)), I could observe reported non-BE for ISM MD files with the only difference being line endings on manual inspection.

Proposal is to switch to a different comparison method for csv files. Reading as text should normalize the line endings