Deprecate old, duplicate functions

[magnunor]

Currently there is duplication of functionality between some of the functions which came from pixstem. One example of this is Diffraction2D.shift_diffraction, which essentially does the same thing as Diffraction2D.center_direct_beam.

These duplicate functions should firstly get a DeprecationWarning, then be removed in a future release.

[pc494]

It's also (as it happens) a special case of Diffraction2D.apply_affine_transformation().

Shall we use this thread as a place to keep track of such duplications?

[magnunor]

Shall we use this thread as a place to keep track of such duplications?

Sounds good!

Some other functions are radial_integration and radial_average.

I'm tempted to wait until #759 (via hyperspy/hyperspy#2703) is complete, so that the "new" functions can be implemented using s.map.

[magnunor]

I've gone through the old pixstem functions in Diffraction2D, seeing what can be removed, and what should be tweaked.

---

I'm tempted to remove some of the the old pixstem functionality which overlaps with other functionality. For example:

• Diffraction2D.shift_diffraction is pretty similar to center_direct_beam
• Diffraction2D.rotate_diffraction, Diffraction2D.flip_diffraction_x and Diffraction2D.flip_diffraction_y can be replaced with two lines of code + s.map. (They're old functions from when HyperSpy's s.map functionality did not handle large datasets very well)
• Diffraction2D.peak_position_refinement_com functionality will be moved into DiffractionVectors?
• Diffraction2D.intensity_peaks functionality will be moved into DiffractionVectors?
• Diffraction2D.make_probe_navigation I think this is implemented directly in HyperSpy now? With the navigator being saved in the metadata?
• Diffraction2D.plot I think all of the functionality here is automatically handled in HyperSpy now?
• Diffraciton2D.angular_mask this is connected to angular_slice_radial_average, which is being deprecated in 1.0.0. As the functionality is replaced with the get_azimuthal_intergral functions

---

Some other functions I think should be kept, but might need some tweaking?

• Diffraction2D.find_dead_pixels I think this is fine
• Diffraction2D.find_hot_pixels I have only ever used this for data acquired on a misconfigured detector. It is different from find_dead_pixels in that the misbehaving pixels can change for each probe position. I'm not sure how often that actually occurs
• Diffraction2D.correct_bad_pixels this can handle the output from both find_dead_pixels and find_hot_pixels, ergo with the mishaving pixels changing at each probe position. This adds a lot of complexity to the function, and I'm not sure how well it performs nowadays.
• Diffraction2D.threshold_and_mask I think this functionality is useful. But due to changes in how center_of_mass is handled in the DPC-restructure pull request, it needs to be updated.

---

• Diffraction2D.template_match_disk preprocessing for some functions, very useful, keep

[pc494]

So I (briefly) went after some of these. There is some discussion in #1019 and #1011.

I think my main feeling was that the namespace on Diffraction2D is so crowded by Signal2D that being 'clean' has a fairly poor effort/reward trade-off.