Add ability to use multiple processes when fitting multiple Roman SCAs

[rmjarvis]

This is the next PR for the Roman PSF. The main feature it adds is to allow multiprocessing during the fits of the 18 SCAs.

We already had a utility function called run_multi, which we use in a few other places as well. So most of the changes here are reorganizing the work for each SCA so there is a self-contained job to pass to each worker.

There are a few other minor cleanup items I'm also including:

• When using hsm_size_reject or other things that use hsm, it didn't reject objects for which hsm gave an error. It should have and now it does.
• In the StarImages output plots, I added the chip number and x,y position to the labels to make it easier to see if there are problems in one or a couple particular SCAs.
• Also for StarImages, if there were fewer stars than requested plots, it didn't work right. I fixed that.
• I added a warning guard for a newish scipy warning that was being emitted sometimes.
• The number of iterations used wasn't being written to the output file. This was an oversight, and it's useful to have that information for diagnostics.

[aeharbo]

New functions and tests look good. Adding the extra blank lines for clarity on input.py is a good move.

Glad you didn't have to deal with too many re-indexing issues outside of the test functions.

[rmjarvis]

Thanks both.