Add HTTP endpoints discovery and overview dashboard

[Makisuo]

Summary

This PR adds a new Endpoints page that displays HTTP server endpoints discovered from traces, with aggregated metrics including latency percentiles, throughput, and error rates.

Key Changes

• New Endpoints Page (/endpoints): Displays a table of HTTP endpoints with filtering by time range and service
• Tinybird Queries: Added two new endpoints to the domain:
  • http_endpoints_overview: Aggregates metrics (count, latency percentiles, error rate) for HTTP server spans grouped by service, method, and route
  • http_endpoints_timeseries: Provides time-bucketed throughput and error rate data for sparkline visualizations
• EndpointsTable Component: Renders a comprehensive table with:
  • HTTP method badges with color coding (GET, POST, PUT, PATCH, DELETE)
  • Endpoint names and service links to related pages
  • Latency metrics (P50, P95, P99) with human-readable formatting
  • Sparkline charts for throughput and error rate trends
  • Formatted count and error rate displays
• Navigation: Added "Endpoints" link to the sidebar under telemetry section with chart bar icon
• Atom Integration: Created query atoms for both endpoints overview and sparklines data with 30-second stale time
• Loading & Error States: Includes skeleton loading state and error boundary handling

Implementation Details

• HTTP endpoints are identified from server spans with SpanKind = 'Server', extracting the route from http.route attribute
• Latency values are converted from nanoseconds to milliseconds for display
• Error rate is calculated as percentage of error status codes
• Sparkline data is filled with zeros for missing time buckets to ensure continuous visualization
• The table supports filtering by service name and time range, with links to drill down into traces and service details

https://claude.ai/code/session_01TvkFU6mqAgG7UFcFf6DMTk