Fix failing unit test in Python 3.13

[mcara]

Fixes #285

CC: @braingram

[braingram]

I cherry picked e1d446d onto #287 so the wheel building would be attempted (which runs the unit tests) with python 3.13 and the change in this PR. Unfortunately the unit test failed in the same way as reported in the issue:
https://github.com/spacetelescope/spherical_geometry/actions/runs/13044906258/job/36393793180?pr=287

Also, I think the CI disabled itself due to inactivity in this repo. I don't have permissions to re-enable it. Do you?
https://github.com/spacetelescope/spherical_geometry/actions/workflows/spherical_geometry.yml

[mcara]

JWST regression tests: https://github.com/spacetelescope/RegressionTests/actions/runs/13045302772

[mcara]

Unfortunately I started regression tests before reading your comment. In my local testing at some moment I switched to numpy 1.26.3 and it made no difference (test was still failing). After this fix all tests were passing. I forgot to test with numpy 2.2.0. I will investigate.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.25%. Comparing base (0f5f0a3) to head (c46e72b).
Report is 2 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #288   +/-   ##
=======================================
  Coverage   80.25%   80.25%           
=======================================
  Files           5        5           
  Lines        1013     1013           
=======================================
  Hits          813      813           
  Misses        200      200           

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

[mcara]

@braingram Try again. I think it should work now but I can't get regression test run past sonar scan.

[braingram]

@braingram Try again. I think it should work now but I can't get regression test run past sonar scan.

Oof, sonar scan is blocked by spacetelescope/jwst#9112 You can use the branch for that PR if you're impatient like me :)

[braingram]

To check I'm understanding. Previously providing np.nan to length resulted in ValueError: Out of domain for acos and with this PR it doesn't raise an exception and instead returns np.nan?

Where is length used and will this change in behavior have consequences?

[mcara]

Another attempt to run regression tests: https://github.com/spacetelescope/RegressionTests/actions/runs/13061746792

[braingram]

LGTM and testing locally this fixes the python 3.13 failure.

To be safe let's wait till the regtests pass then I'll update the 3.13 PR.

Thanks for finding the fix!

[mcara]

Regressiuon tests have numerous failures but they do not appear to be related to this PR but rather to pmap context change and other changes (most non-pmap failing tests are related to MIRI MRS). I tried running another test with jwst_1323.pmap but now tests appear to error instead of failing. If this PR were to affect JWST pipeline, I would expect to see failures in all imaging modes (and not only for MIRI MRS) which I do not see in the regression tests. In fact, I think the failures are similar to failures in other un-related regression tests, e.g., https://github.com/spacetelescope/RegressionTests/actions/runs/13061224788

So, ... merging this