Changes

• feat(dashboard): persist table filters, facets, search, sort and pagination in the url
  • PR: #2798
• test(e2e): cover monitor -> incident -> on-call -> user alerted end to end
  • PR: #2805
• feat(network): site hierarchy, ARP/FDB endpoint discovery, and map drill-down
  • PR: #2795
• revert(e2e): loosen createMonitor URL match to fix master e2e regression
  • PR: #2824
• fix(analytics): parse ClickHouse count() returned as a JSON number — telemetry monitors never went Offline
  • PR: #2826
• Network observability overhaul: one cohesive Network product
  • PR: #2827
• Network observability: trap history, device logs, flow analytics v2, per-interface & anomaly alerting, site alerts, NetFlow v9, IPv6
  • PR: #2829
• feat(kubernetes): Kubernetes cost observability (OpenCost/Kubecost integration)
  • PR: #2828
• docs(i18n): translate Kubernetes cost observability docs into all 15 locales
  • PR: #2831
• feat(dashboard): merge Network Devices and Network Sites into one Network products-menu entry
  • PR: #2833
• feat(network-devices): devices own polling and data collection; monitors become the alerting layer
  • PR: #2835
• fix(analytics): table-qualify aggregate WHERE clauses so SELECT aliases cannot capture them
  • PR: #2836
• feat(network): form steps on long forms, configurable site types
  • PR: #2837
• feat(slo): SLOs, error budgets, burn-rate alerting and an SLI dashboard widget
  • PR: #2838
• fix(kubernetes-cost): make cost.enabled=true work end to end
  • PR: #2840
• fix(network): poll schedule timezone skew, topology framing, and device UX gaps
  • PR: #2839
• chore(dashboard): delete dead DashboardComponents/Index.ts
  • PR: #2841
• feat(kubernetes-agent): diagnose the cost pipeline in troubleshoot.sh
  • PR: #2842
• feat(slo): fill the UI/UX gaps in the SLO/SLI product
  • PR: #2844
• fix(dashboard): render Kubernetes cost breakdowns with the standard Table
  • PR: #2843