Add service observability and rollout metrics

[kacy]

Summary

• Add service observability Prometheus metrics for reconcile activity, BPF sync failures, health latency, endpoint flaps, and L7 proxy state.
• Extend rollout observability via /v1/status?mode=service_rollout and /v1/metrics?format=prometheus.
• Update the user guide to document the rollout observability surfaces and the new per-service metrics.
• Stabilize proxy, steering, listener, registry, and status-metrics tests so the observability branch validates cleanly in this environment.

Validation

• env YOQ_SKIP_SLOW_TESTS=1 ZIG_GLOBAL_CACHE_DIR=.zig-global-cache ZIG_LOCAL_CACHE_DIR=.zig-local-cache zig build test
• Result: 1686 passed; 10 skipped; 0 failed

Review Notes

• Please verify that the Prometheus metric names and labels are acceptable for downstream consumers.
• Please verify that the health-gated endpoint behavior reflected in the status-metrics test is the intended runtime behavior.
• Listener and steering tests were made sandbox-aware so constrained environments skip or behave deterministically instead of failing spuriously.