Add extract_md5sum check

[wholtz]

Checklist

• Pull request details were added to HISTORY.rst

I frequently have workflows that generate *.fastq.gz files as outputs. Because gzip embeds a timestamp in the file, md5sum comparisons fail. gzip with -n can be used to omit the timestamp, but the gzipping is often part of third-party tool and not configurable. This PR adds a ungzip_md5sum option that runs md5sum on the ungzipped contents of the file.

[rhpvorderman]

This is a good idea, but let's prevent the need for unxz_md5sum and unbzip2_md5sum. I opt for extract_md5sum. xopen can be used to transparently open files with all these compressions.

Then the dichotomy becomes:

• md5sum open as is, even if it is an archive and just sum the bytes.
• extract_md5sum extract the archive using xopen and sum those bytes.

That seems pretty maintainable and doable to me. Only minor changes to this PR are needed. Do you want to do it yourself or do you rather want me take it over?

[wholtz]

Thanks for the review @rhpvorderman, I have made the requested modifications. Feel free to tweak as needed.

[rhpvorderman]

Excellent. This is exactly what I had in mind. Great feature idea and implementation. Thanks!