docs: add AI policy. Starting point until the foundation-level

[Marc-cn]

Summary

Adds AI_POLICY.md, adapted from in-toto's. Project names updated, contributor-guidelines link pointed at this repo's CONTRIBUTING.md. Attribution to in-toto retained in the opening line.

Type of Change

• Bug fix (non-breaking change fixing an issue)
• New feature (non-breaking change adding functionality)
• Breaking change (fix or feature causing existing functionality to change)
• [x ] Documentation update
• Refactoring (no functional changes)

Framework Changes Checklist

N/A -no changes under packages/darnit/.

Control/TOML Changes Checklist

N/A - no control or TOML changes.

Testing

N/A - documentation only, no code touched.

Additional Notes

Two things worth deciding before this leaves draft:

1. DCO. The policy requires Signed-off-by, but only 2 of the last 20 commits carry one and there's no DCO check in CI. Either we enable enforcement or soften that section.
2. Scope. This covers AI-assisted contributions. darnit also calls an LLM at runtime as part of the product, which is a different question and isn't addressed here. Might warrant its own section or a separate document.