docs(agents): add security analysis protocol to principal-engineer-reviewer

[johntmyers]

Summary

Adds concrete security analysis guidance to the principal-engineer-reviewer agent, including framework references (CWE, OWASP ASVS, OWASP Top 10 for LLM, CAPEC), a prescriptive threat modeling protocol, and a security checklist. The new guidance activates contextually for security-sensitive reviews only.

Changes

• Enhanced the Security priority with explicit framework references (CWE, OWASP ASVS Level 3, OWASP Top 10 for LLM Applications, CAPEC)
• Added a "Security analysis" section with a 5-step protocol: threat modeling, weakness mapping with CWE IDs, sandbox integrity checks (Landlock/seccomp/YAML policy), input sanitization rules, and dependency auditing
• Added a security checklist covering CWE-78/88, CWE-94, CWE-22, CWE-269, OWASP LLM06, and supply chain risks
• Scoped activation to security-sensitive areas (sandbox runtime, policy engine, network egress, auth, credential handling, untrusted input) to avoid over-applying on non-security PRs

Testing

• Verified file reads correctly and sections flow coherently
• mise run pre-commit passes
• Unit tests added/updated

Checklist

• Follows Conventional Commits
• Commits are signed off (DCO)
• Architecture docs updated (if applicable)