🧠 Enhanced LLM Export
- Historical Context: The export now includes the last 48 hours of events, last 10 speedtests, and active Incidents.
- Mode Selector: Choose between Full Context (for initial diagnosis) and Update (last 6h) to save tokens.
- Why? Allows ChatGPT/Claude to analyze patterns (e.g. "Modulation drops every night at 3 AM") instead of just a single snapshot.
⏱️ Performance
- Optimized DB queries for export generation.