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 <cursoragent@cursor.com>

internal/httpclient/circuit.go

@@ -49,6 +49,24 @@ func (b *hostBreaker) recordFailure() {
 	}
 }
 
+// SnapshotBreakers returns whether each known host breaker is currently open.
+func SnapshotBreakers() map[string]bool {
+	out := map[string]bool{}
+	hostBreakers.Range(func(key, value any) bool {
+		host, _ := key.(string)
+		b, _ := value.(*hostBreaker)
+		if b == nil {
+			return true
+		}
+		b.mu.Lock()
+		open := time.Now().Before(b.openUntil)
+		b.mu.Unlock()
+		out[host] = open
+		return true
+	})
+	return out
+}
+
 // ResetHostBreakers clears all circuit breakers (tests only).
 func ResetHostBreakers() {
 	hostBreakers = sync.Map{}

internal/httpclient/httpclient.go

@@ -7,6 +7,8 @@ import (
 	"io"
 	"net/http"
 	"time"
+
+	"evobgp/internal/observability"
 )
 
 const DefaultTimeout = 45 * time.Second
@@ -69,19 +71,24 @@ func DoWithBreaker(ctx context.Context, hc *http.Client, req *http.Request, maxA
 	if req == nil || req.URL == nil {
 		return nil, fmt.Errorf("httpclient: nil request")
 	}
-	br := breakerForHost(req.URL.Hostname())
+	host := req.URL.Hostname()
+	br := breakerForHost(host)
 	if !br.allow() {
-		return nil, fmt.Errorf("httpclient: circuit open for %s", req.URL.Hostname())
+		observability.SetUpstreamBreakerOpen(host, true)
+		return nil, fmt.Errorf("httpclient: circuit open for %s", host)
 	}
 	resp, err := DoWithRetry(ctx, hc, req, maxAttempts)
 	if err != nil {
 		br.recordFailure()
+		observability.SetUpstreamBreakerOpen(host, !br.allow())
 		return nil, err
 	}
 	if resp.StatusCode >= 500 {
 		br.recordFailure()
+		observability.SetUpstreamBreakerOpen(host, !br.allow())
 		return resp, nil
 	}
 	br.recordSuccess()
+	observability.SetUpstreamBreakerOpen(host, false)
 	return resp, nil
 }