refactor: interp2d now uses grid_sample where possible

[ConnorStoneAstro]

No description provided.

[codecov]

Codecov Report

Attention: Patch coverage is 83.82353% with 11 lines in your changes missing coverage. Please review.

Project coverage is 92.28%. Comparing base (896f7b5) to head (5c4cba7).

Files with missing lines	Patch %	Lines
src/caustics/utils.py	75.00%	10 Missing ⚠️
src/caustics/lenses/pixelated_convergence.py	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #375      +/-   ##
==========================================
- Coverage   93.76%   92.28%   -1.48%     
==========================================
  Files          57       57              
  Lines        2599     2619      +20     
==========================================
- Hits         2437     2417      -20     
- Misses        162      202      +40     

Flag	Coverage Δ
unittests	92.28% <83.82%> (-1.48%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

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

[rouzib]

I went throught my own tests comparing the output of the changed functions as well as their gradient. All my tests remain consistent between the current main branch and the new fast interp2d.

Should we expose the interpolation mode in the lenses to the user? If so, we could add the bicubic mode as an option.

Changes:

1. Removed "extrapolate" and "clamp" as a padding_mode option since "extrapolate" is not supported by the old_interp2d and "clamp" is not supported by grid_sample.
2. Updated interp2d documentation.