Populate sky metadata with zero if matching is not possible.

[schlafly]

Make sure that we send sensible metadata even when the L2 images don't overlap one another. The existing code populates Nones here, which don't fit into the individual_image_metadata tables with the other numbers.

Regression tests are passing: #1355

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78.58%. Comparing base (8d7d7aa) to head (dd006ea).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1360   +/-   ##
=======================================
  Coverage   78.57%   78.58%           
=======================================
  Files         117      117           
  Lines        7861     7862    +1     
=======================================
+ Hits         6177     6178    +1     
  Misses       1684     1684           

Flag	Coverage Δ		*Carryforward flag
nightly	62.26% <ø> (ø)		Carriedforward from 8d7d7aa

*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.

[ddavis-stsci]

This fixes the issue I was seeing,

rerun tests
strun --debug roman_mos r0000101001001001001_r274dp63x32y80_visit_F158_prompt_i2d_asn.json
2024-08-12 10:20:41,812 - stpipe.MosaicPipeline - INFO - MosaicPipeline instance created.
...
2024-08-12 10:32:58,174 - stpipe.MosaicPipeline - INFO - Saved model in r0000101001001001001_r274dp63x32y80_visit_F158_prompt_i2d.asdf
[stpipe.MosaicPipeline] Saved model in r0000101001001001001_r274dp63x32y80_visit_F158_prompt_i2d.asdf
2024-08-12 10:32:58,174 - stpipe.MosaicPipeline - INFO - Step MosaicPipeline done
[stpipe.MosaicPipeline] Step MosaicPipeline done

asdfattr.py r0000101001001001001_r274dp63x32y80_visit_F158_prompt_i2d.asdf cal_step
attributes: cal_step
ex_list: ['cal_step']
Meta Attribute cal_step
flux COMPLETE
outlier_detection COMPLETE
skymatch COMPLETE
resample COMPLETE

Do we need this in for the release. My take is that it would be nice but not needed.

[schlafly]

Regression tests are passing. https://github.com/spacetelescope/RegressionTests/actions/runs/10365669711

It doesn't address a specific requirement and is in that sense probably optional, but I don't see why we wouldn't include it. Likewise I think we should get in #1355, as it fixes a bug, but isn't critical. The latter requires some okify'ing, but only to get rid of the previous problematic extra context planes.