API: Move finite support probe constraints to probe constraints section

[carterbox]

Purpose

Organize the code and make algorithm implementation less complicated by moving the probe constraints all to one section.

Approach

Convert the additional probe penalty and finite support probe constraints to the probe constraint section, and implement them as soft constraints.

Adjust the constrain center peak function to only take single pixel steps. This prevents large and sudden movements of the probe function in order to keep it centered. The method for determining the center is now also center of mass instead of finding the brightest spot.

Pre-Merge Checklists

Submitter

• Write a helpfully descriptive pull request title.
• Organize changes into logically grouped commits with descriptive commit messages.
• Document all new functions.
• Click 'details' on the readthedocs check to view the updated docs.
• Write tests for new functions or explain why they are not needed.
• Address any complaints from pep8speaks.

Reviewer

• Actually read all of the code.
• Run the new code yourself; the included tests should make this easy.
• Write a summary of the changes as you understand them.
• Thank the submitter.

[pep8speaks]

Hello @carterbox! Thanks for opening this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file tests/ptycho/test_probe.py:

Line 180:1: E302 expected 2 blank lines, found 1
Line 184:8: E231 missing whitespace after ','
Line 184:10: E231 missing whitespace after ','

[a4894z]

Looks good, I just had one main question/clarification for the probe.py file

[a4894z]

It looks like the main change for centering the probe is using the center of mass of the probe intensity instead of its max value?

[carterbox]

Yes, that's the main change. However, there are also a few implementation details which I think I mentioned in the commit message?

1. The gaussian filter used to smooth the probe beforehand is now truncated at 6 sigma.
2. The gaussian filter now assumes there is no probe outside the probe array (cval=0.0)
3. The sigma is chosen so 6 sigma is the probe width.
4. The shifts applied to the probe are now only elements from the set [-1, 0, 1]. This keeps the shifts small and avoids interpolation error because all shifts are integer.

[a4894z]

I like this way better with all the probe constraints applied here instead of in the rPIE or LSTSQ code.

Also, at some point I think we can still use Difference Map ideas for applying these probe or sample constraints...I'll show you what I mean when we meet tomorrow afternoon.