feat(learning): add guided waterbodies notebook and clearer lessons

[fabricebrito]

The introductory waterbodies exercise previously asked students to inspect fixture code and run pytest. This change provides a guided notebook where students predict results, view spectral bands and intermediate rasters, classify water, and inspect validated STAC metadata.

• Link the exercise to JupyterLab and VS Code launch instructions, with an optional notebook environment.
• Add AOI, band-order and mismatched-grid experiments using the existing application implementations.
• Explain custom-schema benefits with datetime and URI examples, and introduce Click before describing generated CLIs.
• Record documentation style guidance in AGENTS.md.
• Replace 12 Git-based tooling dependencies with matching exact PyPI release pins and refresh the lockfile.

Validation: required Hatch format, lint, type and security checks passed; 38 tests passed and 1 skipped. Tests also passed against the installed PyPI tooling. The notebook executed successfully for the default scene, both uniform-scene AOIs and reversed bands. The strict MkDocs build passed. Tests used repository-local temporary storage to avoid the host's /tmp quota.