feat(projects): add weather diagnostics toolkit

[stacknil]

Summary

• Add projects/python-weather-diagnostics-toolkit, a public-safe Python mini-lab for ERA5-style weather-field diagnostics.
• Include reusable modules for aliases, thermodynamics, dynamics, regional features, baseline modeling, synthetic fields, and ensemble summaries.
• Add detailed reviewer-facing docs for calculation methods, diagnostic interpretation, data policy, reproducibility, sanitization, and source-to-public mapping.
• Add a scoped CI workflow for the new weather toolkit and a root README entry that keeps sbom-diff-and-risk as the flagship release surface.

Validation

• python -m pip install -e .[dev]
• python -m pytest -> 9 passed
• python -m compileall src scripts
• python -m ruff check .
• CLI smoke checks for run_thermodynamic_check.py, run_dynamics_summary.py, and run_synthetic_ensemble.py
• Synthetic ensemble output generation to ignored outputs/

Privacy and Data Boundary

• No raw weather datasets, NetCDF/GRIB/HDF files, course documents, generated binary figures, compressed archives, or local-machine artifacts are included.
• Public examples are synthetic and labeled synthetic.
• Candidate-file scan returned 0 restricted artifact files, 0 text privacy matches, and 0 non-ASCII text files.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: fffb0709de

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Handle polar latitudes before dividing by dx

When latitude includes ±90° (common in ERA5-style grids), np.cos(np.deg2rad(lat)) becomes zero, so dx is zero and gradient_on_latlon divides by zero, producing inf/nan derivatives at the pole rows. This propagates directly into relative_vorticity and horizontal_advection, corrupting diagnostics for full-domain datasets that include poles. Please mask pole rows, exclude poles, or enforce a minimum nonzero zonal spacing before the division.

Useful? React with 👍 / 👎.

[stacknil]

Addressed the polar-latitude diagnostic comment in follow-up commit 78942f9. Exact pole rows now mask undefined zonal gradients as NaN, with a regression test covering +/-90 degree latitude rows and updated calculation-method notes.