perf: Add performance monitoring and metrics collection

[manavgup]

Problem

No visibility into performance issues:

• Can't track query counts per request
• No alerting on performance degradation
• Difficult to diagnose bottlenecks

Root Cause

• No performance monitoring infrastructure
• No metrics collection
• No dashboards or alerts

Proposed Solution

1. Create monitoring/metrics.py with PerformanceMonitor class
2. Add decorators to track database queries, LLM calls, token usage
3. Log performance summary per request
4. Set up alerts for threshold violations

Expected Impact

• Visibility into performance metrics
• Proactive issue detection
• Data-driven optimization decisions

Files to Create

• backend/monitoring/metrics.py
• backend/monitoring/alerts.py

Files to Modify

• Database query execution points (add tracking decorator)
• LLM provider calls (add token tracking)
• Main request handlers (add summary logging)

Acceptance Criteria

• PerformanceMonitor class implemented
• Metrics tracked: db_query_count, llm_token_count, execution_time
• Alerts trigger on thresholds (> 20 queries, > 5s execution)
• Metrics logged per request
• Dashboard-ready format (JSON structured logs)

Priority

Medium - Phase 3 Advanced Optimization

Related Issues

Part of overall performance optimization initiative