New TIKE functionality for missing measurement regions on detector and Poisson noise model for exitwave update

[a4894z]

Purpose

This pull request adds the following features to TIKE:

1. Uses a binary mask to define unmeasured pixels from a ptycho imaging experiment and then only performs exitwave updates in Fourier space at the measured pixels and leaves unconstrained the unmeasured regions.

2. Introduces the option of using the Poisson noise model for the exitwave update, and performs step length computations using a) the method from ptychoshelves (which only computed steplengths for the dominant exitwave mode and uses that steplength for all the less dominant modes), or b) uses a similar method as ptychoshelves but computing steplengths for each mode individually.

I probably should have split this into two pull requests, but we’re at where we’re at.

Approach

For the unmeasured pixels part, I use a binary array that is loaded from an external source (if not found, we default to assuming no unmeasured pixels) which is then used to define regions where the exitwave gradients are defined and used in the exitwave updates, and regions where the gradient is undefined I leave unconstrained.

For the Poisson noise model exitwave updates, I implement the step length computation as we have discussed during our regular TIKE meetings...in a nutshell: I introduce two methods for computing steplengths for exitwave updates, 1) using the method found in ptychoshelves which only uses the dominant mode for steplength computation and a method I developed which does a vectorized computation of steplength for each exitwave mode separately.

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.

[carterbox]

Thanks for the docstrings! Please respond to the unaddressed comments above.

[a4894z]

see new commit, I think I'm caught up on the issues you raised?

[carterbox]

Now that there's no axis reordering, it's a lot more readable. Thanks!