Enhance LLM reliability with timeout handling and add incident timeline generator

[Mahendrareddy2006]

This PR improves the robustness of the LLM integration and introduces an "Incident Timeline Generator" for structured emergency reports.

Changes:

Enabled 'type_check_all()' validation in 'main_loop()'
Made the Ollama model configurable via 'OLLAMA_MODEL' environment variable
Added timeout to'requests.post()' to prevent hanging
Improved JSON parsing using 'json_data.get()' to avoid crashes
Added missing import for 'Union' in 'main.py'
Added a new timeline_extractor module to extract chronological events from incident narratives
Integrated timeline extraction into the 'Controller' pipeline
Added tests to validate timeline extraction and controller integration
Removed duplicate methods in the controller and cleaned integration logic

These improvements make the system more stable when interacting with Ollama and add support for generating structured incident timelines from narrative reports.
Fixes #286

[Mahendrareddy2006]

Hi! My name is Mahendra Reddy and I’m interested in contributing to FireForm for Google Summer of Code 2026.

I recently started exploring the repository and submitted this PR to improve the reliability of the LLM request handling.

I’m currently learning more about the FireForm architecture and would appreciate any feedback or suggestions on other areas where contributors could help. Thank you!