Use astropy pixel_to_pixel in calc_pixmap

[mcara]

Closes #51

[codecov[bot]]

Codecov Report

:white_check_mark: All modified and coverable lines are covered by tests. :white_check_mark: Project coverage is 97.52%. Comparing base (d6fc5bd) to head (22abb77).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #161      +/-   ##
==========================================
+ Coverage   97.51%   97.52%   +0.01%     
==========================================
  Files           3        3              
  Lines         241      242       +1     
==========================================
+ Hits          235      236       +1     
  Misses          6        6              

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

:rocket: New features to boost your workflow:

• :snowflake: Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[nden]

This will need careful consideration, and as a minimum a regression test run. Looking at the astropy pixel_to_pixel function, it uses the high level API pixel_to_world which generates SkyCoord objects. It's supposed to be efficient with views of arrays. All I'm saying is it needs some more testing specifically for the use in drizzle.

[mcara]

There are unit tests testing calc_pixmap in drizzle. calc_pixmap is not used in the pipelines and so running regression tests would not test this helper function.

[braingram]

There are unit tests testing calc_pixmap in drizzle. calc_pixmap is not used in the pipelines and so running regression tests would not test this helper function.

How would you like to coordinate this with https://github.com/spacetelescope/stcal/pull/320 (which will start using calc_pixmap in stcal and the pipelines)?

[mcara]

This cannot be merged because after https://github.com/spacetelescope/jwst/pull/9404 NIRSPEC WCS no longer works with separability_matrix - see https://github.com/astropy/astropy/issues/18523.

So we either wait for separability_matrix() to be fixed or modify NIRSPEC WCS to propagate name in a different way that doesn't crash separability_matrix() used by pixel_to_pixel().