feat(jobs): add durable PG queue reclaim, slog, and richer metrics

JSON slog в ключевых пакетах; Prometheus path_group, job_audit_depth, upstream breaker; job_audit ClaimQueued/ReclaimStaleRunning + Adopt loop для HA после рестарта.

Co-authored-by: Cursor <[email protected]>

internal/httpapi/server.go

@@ -14,6 +14,7 @@ import (
 	"evobgp/internal/jobs"
 	"evobgp/internal/maintenance"
 	"evobgp/internal/pgmonitor"
+	"evobgp/internal/repository"
 	"evobgp/internal/runtimelogs"
 	"evobgp/internal/store"
 
@@ -156,6 +157,9 @@ func (s *Server) Jobs() *jobs.Registry { return s.jobs }
 func (s *Server) StartBackground(ctx context.Context) {
 	if s != nil && s.pgPool != nil {
 		pgmonitor.StartScheduler(ctx, s.pgPool)
+		if s.jobs != nil {
+			jobs.StartDurableQueueLoop(ctx, s.jobs, repository.NewJobAuditWriter(s.pgPool))
+		}
 	}
 	if s != nil && s.maintConfig != nil && s.jobs != nil {
 		maintenance.StartScheduler(ctx, s.maintConfig, func(policyID string, dryRun bool, idem string) {