Spatially-variant PSF/background reproducability

[aakashdp6548]

Mostly this PR is to reproduce the results and figures from the spatially-variant PSF paper.

I've made some changes to metrics, including:

• Changing SourceTypeAccuracy to bin by magnitude instead of flux for consistency with other metrics
• Added binning by magnitude for FluxError and GalaxyShapeError

[aakashdp6548]

Hey @jeff-regier, here is the code to reproduce the results/figures for the PSF paper. I haven't pulled from main yet so it's not ready to merge, just wanted to get this up as a draft so I don't forget.

I made some changes to the metrics (mentioned in the first post). Can you let me know if these are fine to change, or if they may break someone else's results? I can create a new metric instead but didn't want to increase code bloat...

[codecov]

Codecov Report

Attention: Patch coverage is 94.07407% with 8 lines in your changes missing coverage. Please review.

Project coverage is 82.13%. Comparing base (c0645a3) to head (c795a3f).

Files	Patch %	Lines
bliss/__init__.py	12.50%	7 Missing ⚠️
bliss/encoder/metrics.py	99.20%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1053      +/-   ##
==========================================
+ Coverage   81.03%   82.13%   +1.10%     
==========================================
  Files          23       23              
  Lines        2873     2978     +105     
==========================================
+ Hits         2328     2446     +118     
+ Misses        545      532      -13     

Flag	Coverage Δ
unittests	82.13% <94.07%> (+1.10%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jeff-regier]

Thanks Aakash!