feat: Add OPEN_AI_LLM_TIMEOUT environment variable for LLM timeout

[Jah-yee]

Summary

• Adds OPEN_AI_LLM_TIMEOUT environment variable to configure LLM timeout
• Default is 900s (existing behavior) if not set
• Allows users to override when using local LLM providers like Ollama or LM Studio

Changes

• run_agent.py: Changed hardcoded timeout to use os.getenv("OPEN_AI_LLM_TIMEOUT", 900.0))

Testing

• Existing tests should pass (no behavior change when env var is not set)
• Can be tested by setting OPEN_AI_LLM_TIMEOUT=1800 for 30-minute timeout

Closes #1010

[teknium1]

Thanks for the PR! This has already been implemented as HERMES_API_TIMEOUT (same default of 900s) in PR #1194 — we went with a provider-agnostic name since it applies to all backends, not just OpenAI. Closing as superseded.