adding average_radius parameter into the rotation wrapper

[dkazanc]

This PR should be used together with the vocentering branch of httomolibgpu.
The changes include adding a new parameter average_radius into the rotation wrapper in order to perform averaging of multiple sinograms to improve SNR and therefore CoR estimation accuracy.

Checklist

• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I have made corresponding changes to the documentation

[yousefmoazzam]

After mulling this over for a bit, I think there could be a way to test this more thoroughly.

If I understand the added logic correctly (and please do correct me if I've got things wrong), there are 4 cases:

1. the average_radius parameter isn't provided to the method; the method wrapper sets average_radius=0, and thus the underlying method function should be given a non-averaged sinogram
2. the average_radius parameter is provided to the method

• the value is 0: the underlying method function should be given a non-averaged sinogram
• the value is non-zero but within range: the underlying method function should be given an averaged sinogram, based on the block data and the value of average_radius
• the value is non-zero and out of range: an error should be raised

The last case seems to be tested, but the other 3 cases aren't.

A non-averaged sinogram vs. an averaged sinogram that came from averaging n specific sinograms are distinguishable (ie, they should produce the exact same array). To test the other three cases, I think that one could write the dummy method function that is imported to assert on the array that is passed to it, like this test for example:

httomo/tests/method_wrappers/test_generic.py

Lines 363 to 379 in ebd18a6

	class FakeModule:
	def fake_method(array):
	np.testing.assert_array_equal(array, 42)
	return array
	mocker.patch(
	"httomo.method_wrappers.generic.import_module", return_value=FakeModule
	)
	wrp = make_method_wrapper(
	make_mock_repo(mocker),
	"mocked_module_path",
	"fake_method",
	MPI.COMM_WORLD,
	make_mock_preview_config(mocker),
	)
	dummy_block.data[:] = 42
	wrp.execute(dummy_block)

For each test case, one could determine the sinogram that is expected to be extracted by the wrapper (non-averaged or averaged depending on the value of average_radius) and assert that the method function gets given this expected sinogram when the wrapper is called.

[dkazanc]

@yousefmoazzam I've added two more tests, as you suggested above, so those situations are covered now :

1. the value is 0: the underlying method function should be given a non-averaged sinogram
2. the value is non-zero but within range: the underlying method function should be given an averaged sinogram, based on the block data and the value of average_radius
3. the value is non-zero and out of range: an error should be raised

[yousefmoazzam]

Thanks for the updates!

Don't worry about IRIS CI failing, it's the flakey cupy import error once more... I ran the tests locally and they're all passing, so that's fine.

I made a few tweaks to the tests that I think makes them a bit clearer, where the first commit's message explains the changes. I also added a test for the last missing case (number 1. in my post above) where I think we want the method wrapper to implicitly set the average_radius to 0 if a value for that parameter isn't passed in.

Let me know if what I changed/added makes sense to you (am happy to rollback what I did if there's issues).

[dkazanc]

thanks @yousefmoazzam , yes the changes make sense and the new test as well. I'll merge the PR and do the release of the backends as well.