Bug: 3 intent context prompt files lack doc-first safeguard added in #1476 #1517

Problem

installation_help.md was updated in #1476 to include a prominent CRITICAL: Read Actual Docs Before Answering warning requiring AutoBot to run find/cat commands to read actual documentation before using the static content.

The other 3 intent context prompt files injected by chat_intent_detector.py have the same stale-data risk but do NOT have this safeguard:

• resources/prompts/chat/architecture_explanation.md — static VM architecture description with hardcoded IPs/versions
• resources/prompts/chat/troubleshooting.md — static troubleshooting steps
• resources/prompts/chat/api_documentation.md — static API endpoint list

Impact

AutoBot can still hallucinate outdated architecture details, troubleshooting steps, or API endpoints if RAG retrieval fails to return relevant results for these intents.

Fix

Add the same ## CRITICAL: Read Actual Docs Before Answering block to each file, pointing to the relevant actual documentation files (architecture docs, troubleshooting guide, API reference).

Discovered During

Implementing #1476 and #1494.