2362 update operation window value warnings

[RasmiaKulan]

Issue Closes #2362

Description

• Renamed enable_nonpositive_check to enable_value_check and updated the function to do warning checks specific to zero, negative and NaN values.

• Updated the overlay colors to be user friendly:

  • Zero values

    

  • Negative values

    

  • Invalid values: Not a number

    

• Updated the test_negative_nan_overlay test to check against value warning messages.

• Replaced all the references of enable_nonpositive_check to enable_value_check in the code base.

Developer Testing

• I have verified unit tests pass locally: python -m unittest -v mantidimaging.eyes_tests.reconstruct_window_test

Acceptance Criteria and Reviewer Testing

• Unit tests pass locally: python -m pytest -vs
• Run python -m unittest -v mantidimaging.eyes_tests.reconstruct_window_test and verify all tests pass

Documentation and Additional Notes

• Screenshot tests have been updated

[JackEAllen]

This PR introduces changes to the Applitools screenshot tests

[coveralls]

coverage: 70.805% (-0.02%) from 70.829%
when pulling 237776e on 2362-update_operation_window_value_warnings
into 50f905b on main.

[JackEAllen]

Thank you for adding more clarity to the value warnings. The changes look good and are cleanly implemented. I have functionally tested with real data and ran tests locally and all works well. I have left one small suggestion I would like to get your opinion on, but I don't feel too strongly for the suggested change.

[samtygier-stfc]

I've just given this a quick look and have some questions.

• Is it needed for VALUE_CHECKS to include a nan check? I think all the places that previously used enable_nonpositive_check() also had enable_nan_check(), so that is already being checked.
• Is the conversion to float needed?
• Do _is_zero() and _is_negative() need internal nan checks? Those seem redundant, if a value is zero we already know it is not nan.

Also is there any performance hit. I think with the extra conversions and nan checks there might be.

You could do

    def check_for_bad_data(self) -> None:
        with ExecutionTimer(msg="check_for_bad_data()"):
            current_slice = self._get_current_slice()
            if current_slice is not None:
                for test in self.enabled_checks.values():
                    test.do_check(current_slice)

To get the run time of the checks and switch it to ExecutionProfiler to see where the time is spent. (You might need to turn on preformance logging in the settings)

[samtygier-stfc]

Working well locally for me. Thanks for the fixes.

Requested changes by Jack and I have been made, and I don't think Jack had any further comments. (If there are any those can be resolved when he's back)