Fix scipy 1.14 compatibility and trapz usage

[djhoese]

CC @arcanerr

The scipy.integrate.trapz function is deprecated. This adds a try/except around importing trapezoid versus trapz. I'm not sure how long we want to keep this workaround as trapezoid has been around since at least January of 2023 from what I can tell in git.

• Closes Pyspectral 0.13.1 not compatible with SciPy 1.14.x -> update dependencies #227
• Tests added
• Tests passed: Passes pytest pyspectral
• Passes flake8 pyspectral
• Fully documented

[codecov]

Codecov Report

Attention: Patch coverage is 93.10345% with 2 lines in your changes missing coverage. Please review.

Project coverage is 89.82%. Comparing base (5853de5) to head (239c410).
Report is 43 commits behind head on main.

Files with missing lines	Patch %	Lines
pyspectral/radiance_tb_conversion.py	66.66%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #228      +/-   ##
==========================================
- Coverage   90.00%   89.82%   -0.19%     
==========================================
  Files          23       23              
  Lines        2571     2535      -36     
==========================================
- Hits         2314     2277      -37     
- Misses        257      258       +1     

Flag	Coverage Δ
unittests	89.82% <93.10%> (-0.19%)	⬇️

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.

[djhoese]

FYI for you @adybbroe when you find time to review this: A lot of the test changes are because numpy 2's np.testing.assert_allclose now checks dtype and asserts that dtypes are the same (or at least as far as I can tell). So a lot of the numpy-based tests would use 64-bit float input arrays and then compare against a 32-bit float expected result array. This used to work fine, but now numpy 2 compares about comparing 32-bit with 64-bit. Otherwise, there are just a few spots where numpy 2 now upcasts things where it didn't before.

[adybbroe]

LGTM. Nice with the more compact tests!
I was actually not aware that trapz was deprecated. I think we keep the try-except for now.