fix: decode numpy arrays; feature: topostats versions

[ns-rse]

Decode numpy arrays of strings

I found whilst working through TopoStats refactoring that the inclusion of the config as an attribute to TopoStats objects resulted in some of the values, which are lists of strings, being converted to Numpy arrays (technically everything is converted to Numpy arrays as HDF5 can't handled "lists" of mixed types and so they are coerced to Numpy arrays with a single dtype).

However, these could not be decoded directly and item[()].decode("utf-8") failed with an AttributeError stating that numpy.ndarray does not have attribute decode.

The solution proposed here is to capture this error and if item[()] is an instance of np.ndarray to iterate over the list decoding each item in turn. The typing is explicitly ignored because we want it as a list rather than a dictionary.

Test included, and whilst it passes seems a bit light but does mirror the scenario encountered (had to use group_path as the higher level of nesting).

Handle newer topostats file versions

Restructuring of TopoStats to define classes means the TopoStats object now holds the topostats_version rather than topostats_file_version.

This commit allows both to be handled and switches to using packging.version to do so which ensures a more consistent approach to comparing version numbers.

Yet to write a test for working with newer .topostats where topostats_version >= 2.3.2 as work is still on-going but parameterised test is in place for when work is complete.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 63.63636% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 79.46%. Comparing base (ac9753c) to head (0b1c527).
⚠️ Report is 238 commits behind head on main.

Files with missing lines	Patch %	Lines
AFMReader/io.py	42.85%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #177      +/-   ##
==========================================
+ Coverage   74.62%   79.46%   +4.83%     
==========================================
  Files           8       12       +4     
  Lines         607      935     +328     
==========================================
+ Hits          453      743     +290     
- Misses        154      192      +38     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[ns-rse]

Discovered a minor problem with this as I'd used >= 0.2 when I should have used > 0.2 so have corrected that.

Ready for review and will need merging and releasing around the same time as TopoStats.

[ns-rse]

We could do with merging this at least as the TopoStats Refactor Pull Request (see 1284) requires this fix.

I can work around it for now by installing AFMReader from the branch, but that will be a blocker when we come to release TopoStats v2.4.0 as you can not publish to PyPI packages that require development versions of dependencies.

[SylviaWhittle]

Sorry for the delay. I don't have time to try the code out, but recognise this needs to get merged.