Create data policy and dataset hygiene rules

[rmems]

Goal

Document what Operation Prometheus is allowed to extract, store, and train on.

Context

This project should use public engineering artifacts responsibly and keep sensitive material out of committed datasets.

Tasks

• Add docs/data-policy.md.
• Define allowed public sources:
  • PR metadata
  • issue and PR comments
  • review comments
  • diffs and commits
  • CI/check summaries
  • safe experiment artifacts
• Define excluded sources:
  • private credentials
  • private local configuration
  • raw closed-model chat logs as the primary corpus
  • large model artifacts committed directly to Git
• Add a required manual inspection step before publishing generated datasets.
• Add a rule that raw exports should be treated as temporary working artifacts until normalized and reviewed.

Acceptance Criteria

• docs/data-policy.md exists.
• README links to the policy.
• Policy explains the difference between public engineering history and raw chat-output scraping.