Fix bug when setting meta.background in SkyMatchStep.

[mairanteodoro]

Resolves RCAL-837

This PR addresses a bug in SkyMatchStep that was causing meta.background.subtracted to always be set to None instead of True/False, which was preventing ResampleStep from properly using the sky level as determined by SkyMatchStep and set in meta.background.level.

The changes to the code also required some refactoring to the unit tests.

The code changes in this PR are also related to this old issue: spacetelescope/rad#247

Regression tests
The only failed test has nothing to do with the changes in this PR.
https://plwishmaster.stsci.edu:8081/job/RT/job/Roman-Developers-Pull-Requests/764/
https://plwishmaster.stsci.edu:8081/job/RT/job/Roman-Developers-Pull-Requests/770/
https://plwishmaster.stsci.edu:8081/job/RT/job/Roman-Developers-Pull-Requests/810/

Checklist

• added entry in CHANGES.rst under the corresponding subsection
• updated relevant tests
• updated relevant documentation
• updated relevant milestone(s)
• added relevant label(s)
• ran regression tests, post a link to the Jenkins job below. How to run regression tests on a PR

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78.87%. Comparing base (d10f06b) to head (a8015c0).
Report is 195 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1233   +/-   ##
=======================================
  Coverage   78.87%   78.87%           
=======================================
  Files         117      117           
  Lines        8071     8071           
=======================================
  Hits         6366     6366           
  Misses       1705     1705           

Flag	Coverage Δ		*Carryforward flag
nightly	62.79% <ø> (ø)		Carriedforward from d10f06b

*This pull request uses carry forward flags. Click here to find out more.

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

[bmorris3]

Looks good, it was time for this update. Thanks @mairanteodoro!

[mairanteodoro]

Attaching comparison of the relevant arrays between the output from the regression test against the truth file from Artifactory. Black and white circles indicate the position of the worst absolute and the worst fractional differences, respectively, as reported by the regtest (ran locally, which, by the way, differs from the coordinates reported when ran on Jenkins).

• data array:
  

• err array:
  

• var_flat array:
  

• var_poisson array:
  

• var_rnoise array:

  • worst absolute and fractional differences occurs at the same coordinates
    

• weight array:
  

[schlafly]

Looks good. My rule is that I use "is" when I am comparing objects that may be the same object, and == when I am comparing values. In this context I was surprised by using is to compare bools on 336 but otherwise looks good.

[mairanteodoro]

Looks good. My rule is that I use "is" when I am comparing objects that may be the same object, and == when I am comparing values. In this context I was surprised by using is to compare bools on 336 but otherwise looks good.

Fixed it.
Thanks for catching that overlook of mine. ;)

[nden]

@mairanteodoro This is not attached to a milestone. Was this work included in the last release?

[mairanteodoro]

@mairanteodoro This is not attached to a milestone. Was this work included in the last release?

@nden Not yet. We can see it in this comparison between the main branch and the 0.15.1 release branch (commit 4db6c23): 0.15.1...main

[nden]

@mairanteodoro OK, I added a milestone.