docs: add AI Traceability & Attribution guide

Comprehensive documentation on AI code attribution and disclosure:

- New guide: guide/ai-traceability.md (~640 lines)
  - LLVM "Human-in-the-Loop" policy (Assisted-by trailer)
  - Ghostty mandatory disclosure pattern
  - Fedora contributor accountability framework
  - git-ai tool documentation
  - PromptPwnd security vulnerability
  - Four-level disclosure spectrum
  - Implementation guides (solo, team, enterprise)

- Templates: examples/config/
  - CONTRIBUTING-ai-disclosure.md
  - PULL_REQUEST_TEMPLATE-ai.md

- Cross-references added to:
  - ultimate-guide.md (after Co-Authored-By section)
  - learning-with-ai.md (after Vibe Coding Trap)
  - security-hardening.md (See Also)
  - guide/README.md (table of contents)

- reference.yaml: 14 new entries for AI traceability topics

Source: Vibe coding needs git blame (Piotr Migdał, Jan 2026)
+ Perplexity research on LLVM, Ghostty, Fedora policies

Co-Authored-By: Claude <noreply@anthropic.com>

guide/learning-with-ai.md

@@ -82,6 +82,8 @@ The struggle isn't optional. It's where learning happens.
 
 Term coined by [Andrej Karpathy](https://x.com/karpathy/status/1886192184808149383) (Feb 2025, Collins Word of the Year 2025): coding by "fully giving in to the vibes" without understanding the generated code.
 
+> **Related**: For team and OSS contexts, see [AI Traceability](./ai-traceability.md) for disclosure policies (LLVM, Ghostty, Fedora) and attribution tools.
+
 **Symptoms:**
 - Accept All without reading diffs
 - Copy-paste errors without understanding root cause