Use log files in testing

[lukaspie]

This PR changes the pynxtools.testing framework such that it compares the read_nexus logs directly during the tests. This avoids having to store binary NeXus files in the repos. Instead, we only commit reference logs, which will lead to much smaller diffs and keep the repositories lighter.

For pynxtools-xps, I have a corresponding branch where this change in the testing is already implemented, including a script to autogenerate these log file consistently: https://github.com/FAIRmat-NFDI/pynxtools-xps/pull/69/files#diff-1ac04a8e7b33c27f6b8eea0d6f44313388de34a688758042f1ee9582f5b34ca2. This branch is also installed in the test_plugins pipeline here.

[RubelMozumder]

This is a good idea to prevent our plugin from unnecessarily weighing. We also have to replace nexus files by its log files.
Readme file also needs to updated.

[RubelMozumder]

I had an issue at the very beginning in FAIRmat with the log files, it came from float number precision related to hardware. The float numbers array came with different precision from different machines. I do not know how it works now. Do you think, we can also keep the test option with nexus_file, but we shall recommend using the log_file?

[lukaspie]

I had an issue at the very beginning in FAIRmat with the log files, it came from float number precision related to hardware. The float numbers array came with different precision from different machines. I do not know how it works now. Do you think, we can also keep the test option with nexus_file, but we shall recommend using the log_file?

I believe I made a change in the read_nexus code that fixes this issue at some point, but I can't find it anymore. In any case, we should just fix it there (e.g., by forcing it to always use the same precisison). Can you try if this problem persists?

I am not a big fan of allowing the NeXus file as an alternative, because a) there should be one single way to do it, and b) there is a reason for not having the NeXus files - they are binary and thus there is not git difference, leading to unneccessary bloat of the plugin repos.

[mkuehbach]

Reasonable overall to move forward, but treatment of numeric edge cases is discussion-worthy, branch in need of sync with main, CI fixes

[mkuehbach]

Thanks for this addition, can be merged once updated by main and CI made passing.