@
iostrym - thanks for doing the tests. Sadly I think I'm out of ideas.
If it's any consolation there have been a number of queries recently that could have been addressed more easily if metadata could be extracted from the file name via a regular expression AND the from within the format file; where the file name extraction would take precedence.
If you want to make a formal request for such an enhancement you can do that here ==>>
Bugs : calibre. If you do that then it might useful to include a link to this thread.
I can't assess if such an enhancement is feasible. However, when a CBZ is added the textual metadata can only be extracted from the file name, but the cover is always extracted from the file (1st image). So the rudiments would appear to be there.
Re your last point - its probably unrealistic to expect calibre metadata to be written by anything but calibre - unless you want to do it by hand in a text editor
BR