refactor contour_map function and update test cases with image_comparison (as per matplotlib test guidelines)

[xefonon]

This PR refactors (slightly) the contour_map function in spatial.py and updates the test cases in test_plots.py to use the image_comparison decorator as per Matplotlib test guidelines (https://matplotlib.org/devdocs/devel/testing.html).

Changes Made:
Refactored contour_map Function:

Replaced ax.contour with ax.contourf to prevent breaking and ensure similar-ish results. This solves #91.
Updated Test Cases:

Added @image_comparison decorator to relevant test functions in test_plots.py to validate the visual output against baseline images. This solves (?) #95
Refactored imports to follow Ruff (i.e., imports first, then comment)
Affected Files:
spatial.py
test_plots.py

[xefonon]

OK so apparently my version of matplotlib is not the same as circleci's, so the images compared are "not the same" I guess.

[f-brinkmann]

@xefonon I see that the tests are failing because of the image comparison. We've had problems with that in pyfar as well, where the images generated by matplotlib differed depending on the operation system. https://matplotlib.org/stable/api/testing_api.html#matplotlib.testing.decorators.image_comparison allows to set a tolerance, but the errors we get in testing seem to be quite large. We have RMS values between 7 and 10. Matplotlib gives 0.06 as an RMS error due to numerical precision.

Can you maybe check the baseline to see if everything is correct. Not sure if image comparison makes sense here.

[xefonon]

@xefonon I see that the tests are failing because of the image comparison. We've had problems with that in pyfar as well, where the images generated by matplotlib differed depending on the operation system. https://matplotlib.org/stable/api/testing_api.html#matplotlib.testing.decorators.image_comparison allows to set a tolerance, but the errors we get in testing seem to be quite large. We have RMS values between 7 and 10. Matplotlib gives 0.06 as an RMS error due to numerical precision.

Can you maybe check the baseline to see if everything is correct. Not sure if image comparison makes sense here.

I mean, I created the baseline, so I guess the results can vary depending on the version of one's OS significantly. If no image comparison then what should we do?

[f-brinkmann]

I mean, I created the baseline, so I guess the results can vary depending on the version of one's OS significantly. If no image comparison then what should we do?

In pyfar we still generate images but inspect the changed images manually. That would be fine for me here as well. @mberz do you agree or should we try something else?