Run pytest on macos in GH actions

[reuterbal]

With people using Loki locally on their MacBooks, it seems useful to test this as part of the CI.

[github-actions]

Documentation for this branch can be viewed at https://sites.ecmwf.int/docs/loki/262/index.html

[codecov]

Codecov Report

Attention: Patch coverage is 91.66667% with 3 lines in your changes missing coverage. Please review.

Project coverage is 95.35%. Comparing base (2b3b206) to head (3e41177).

Files	Patch %	Lines
loki/tools/tests/test_tools.py	50.00%	2 Missing ⚠️
loki/tests/test_cmake.py	95.45%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #262   +/-   ##
=======================================
  Coverage   95.35%   95.35%           
=======================================
  Files         171      171           
  Lines       36801    36805    +4     
=======================================
+ Hits        35090    35094    +4     
  Misses       1711     1711           

Flag	Coverage Δ
lint_rules	96.39% <ø> (ø)
loki	95.33% <91.66%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[reuterbal]

I think this is ready for a look. I had to do a few things in addition to ensure compatibility with MacOS, but I think it generally helps to increase being platform-agnostic.

In detail:

• Improved the CMake recipes, using CMake-internal utilities for the Python environment hacking instead of relying on GNU sed syntax
• CMake install now ensures it picks a Python version that is supported
• Added installation instructions for MacOS and updated them for Linux, too
• I had to add a bit more tolerance in the timeout test
• I had to disable test_transform_normalize_array_shape_and_access on MacOS. This looks like it might be a genuine memory corruption problem that only triggers on MacOS - documented in Suspected memory corruption in test_transform_normalize_array_shape_and_access #352 for further investigation (but orthogonal to this PR, I think).

[mlange05]

Looks good to me, and very useful in general. Thanks for grinding this one out! 🙏

Only question I have a is about the tuple-comparison changes, but otherwise this looks great. GTG from me once that is answered.

[mlange05]

🤨 I'd be curious to know why this is necessary?

[reuterbal]

Pytest doesn't like is comparison against literals:

 /home/runner/work/loki/loki/loki/batch/tests/test_batch.py:252: SyntaxWarning: "is" with a literal. Did you mean "=="?
  assert item.dependencies is ()

See for example here: https://github.com/ecmwf-ifs/loki/actions/runs/10124810197/job/27999587354#step:8:120

[mlange05]

Ok, good to know. Thanks.