Add explicit semi-minor axes for spherical projections

[floze-the-genius]

Summary

• complete projection dictionaries that specify only a by copying them and adding an explicit matching b
• preserve projections that already define an ellipsoid or sphere through b, R, ellps, datum, rf, f, es, or e
• update the shipped YAML and legacy area fixtures so their spherical definitions are explicit
• add regression coverage for normalization, input immutability, the create_area_def integration path, and existing ellipsoid parameters
• keep the Python 3.11 mypy analysis compatible with current dependency drift by running its isolated hook on Python 3.12 with numpy<2.3

This keeps the historical spherical interpretation while ensuring stricter CRS consumers receive a complete ellipsoid definition. The mypy compatibility constraint is limited to pre-commit's isolated environment and does not change pyresample's runtime dependencies or Python 3.11 analysis target.

Verification

• JOBS=8 .venv/bin/pytest -q pyresample/test/test_area_config.py (22 passed)

• JOBS=8 .venv/bin/pytest -q pyresample/test (1166 passed, 24 skipped, 2 subtests passed)

• .venv/bin/ruff check pyresample/area_config.py pyresample/test/test_area_config.py

• isolated pre-commit run --all-files (Ruff, whitespace, EOF, YAML, Bandit, mypy, isort)

• git diff --check

• Closes Fix incomplete areas in sample area configurations #200

• Tests added

• Tests passed

• Fully documented

AI disclosure: OpenAI Codex assisted with investigation, implementation, and test execution; the resulting diff and test outputs were reviewed before submission.

[djhoese]

Thanks for creating this PR. It brings up an odd issue and I'm not sure how I feel about fixing it. The description you made talks about "stricter CRS consumers", but do you have any examples of what/who those are? I hesitate to include this in pyresample as it feels like we are defining/forcing defaults that the PROJ C library may be filling in on its own. I do realize PROJ is probably always defining it as you have here.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.69%. Comparing base (2e7a3c8) to head (d70af9f).
⚠️ Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #735      +/-   ##
==========================================
+ Coverage   93.68%   93.69%   +0.01%     
==========================================
  Files          89       89              
  Lines       13715    13743      +28     
==========================================
+ Hits        12849    12877      +28     
  Misses        866      866              

Flag	Coverage Δ
unittests	93.69% <100.00%> (+0.01%)	⬆️

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

☔ View full report in Codecov by Harness.
📢 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.

[floze-the-genius]

You are right to question this. The concrete consumer I had in mind came from the original 2019 report in #200: Satpy GeoTIFF export through rasterio/GDAL failed even though PROJ accepted the projection. I re-ran that path against the current environment here (pyproj 3.7.2 and rasterio 1.5.0): both the a-only and explicit a/b definitions normalize to the same +R=6371228 CRS (EPSG:3409), and both write and reopen a GeoTIFF successfully.\n\nSo I do not have a current stricter consumer that justifies adding runtime normalization in create_area_def; that part of this PR is broader than the evidence supports and duplicates behavior PROJ already provides. I am closing this rather than forcing a default based on a historical failure. If a fixture/docs cleanup is still wanted independently, it would be better handled as a narrower change with current acceptance criteria.

[djhoese]

Thanks for the clarification. In the future, while I can understand if the English language is a barrier, it would be nice to have a conversation with a person rather than an AI.