JP-3664: Numpy 2.0 exposed implicit type promotion

[tapastro]

Partially addresses JP-3664

Closes jwst/#8578
Partially addresses jwst/#8580

This PR addresses a bug exposed by testing numpy 2.0 - the irs2 datamodel specifies arrays to be of type uint8, but operations in refpix promoted them to uint16 implicitly in numpy<2.0. Numpy 2.0 does not implicitly type promote in this case, leading to an integer overflow. This PR updates the specified array datatype to uint16.

Checklist

• added entry in CHANGES.rst (either in Bug Fixes or Changes to API)
• updated relevant tests
• updated relevant documentation
• updated relevant milestone(s)
• added relevant label(s)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 66.18%. Comparing base (84279cf) to head (55ce7e8).
Report is 5 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #319   +/-   ##
=======================================
  Coverage   66.18%   66.18%           
=======================================
  Files         100      100           
  Lines        5397     5397           
=======================================
  Hits         3572     3572           
  Misses       1825     1825           

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

[braingram]

Thanks! The changes look good to me (although a regtest run to be safe seems worthwhile).

I tested saving an irs2 model (with uint8 for OUTPUT and ODD_EVEN) and loading with the new schema. The data is safely cast to uint16 on load (in numpy <2 and 2.1).

[braingram]

I also suggest we leave spacetelescope/jwst#8580 open as this PR doesn't fully address all the numpy 2.0 issues with jwst.

[tapastro]

Agreed on keeping the issue open, and I'll get a regtest run soon. I have an stcal PR out as well, and I suppose I'll need a JWST PR to get the numpy upper limit out of the way.

[tapastro]

Started a regression test run here:

EDIT: make that https://plwishmaster.stsci.edu:8081/job/RT/job/JWST-Developers-Pull-Requests/1674/

[tapastro]

Last set of regtests were failing due to a day of Jenkins issues. Started a new set against jwst/master to check for compatibility issues with numpy<2: https://plwishmaster.stsci.edu:8081/job/RT/job/JWST-Developers-Pull-Requests/1691/