fix: cover runtime LLM API governance patterns

[CoderDeltaLAN]

Summary

• Add runtime governance patterns for tool/CLI dependence on OpenAI, Anthropic, Claude, Gemini, ChatGPT, LLM, external, and remote APIs.
• Extend GOV005 positive coverage for runtime API dependency wording.
• Add a negative case for explicit no-OpenAI-API runtime guidance.

Checks

• ./scripts/check.sh
• workflow_dispatch CI on branch: 27624785043 success

Scope

• src/agent_rules_kit/governance.py
• tests/test_governance.py