fix(cluster): 修复身份事务阻塞与 Worker 容量抖动

将身份协调和安全事件心跳改为 PostgreSQL 单 Leader 执行，并从独立 Worker 进程中移除身份运行时，避免多副本重复写同一状态。

为安全事件事务增加锁等待、空闲事务超时及独立回滚上下文；Worker 需连续丢失六次心跳后才判定失效，降低跨节点抖动导致的容量反复扩缩。

验证：Go 全量测试、go vet、Race 聚焦测试、PostgreSQL 18 Leader/安全事件/Worker 分配集成测试及迁移测试通过。

apps/api/internal/store/worker_registry.go

@@ -9,15 +9,16 @@ import (
 	"github.com/jackc/pgx/v5"
 )
 
-const workerHeartbeatStaleAfter = 15 * time.Second
+const workerHeartbeatStaleAfter = 30 * time.Second
 
 type WorkerRegistrationInput struct {
-	InstanceID      string
-	PodUID          string
-	PodName         string
-	Site            string
-	Revision        string
-	DesiredCapacity int
+	InstanceID          string
+	PodUID              string
+	PodName             string
+	Site                string
+	Revision            string
+	DesiredCapacity     int
+	HeartbeatStaleAfter time.Duration
 }
 
 type WorkerAllocation struct {
@@ -36,6 +37,10 @@ func (s *Store) RegisterWorkerInstance(ctx context.Context, input WorkerRegistra
 	if input.DesiredCapacity < 0 {
 		return WorkerAllocation{}, errors.New("worker desired capacity cannot be negative")
 	}
+	staleAfter := input.HeartbeatStaleAfter
+	if staleAfter < workerHeartbeatStaleAfter {
+		staleAfter = workerHeartbeatStaleAfter
+	}
 	tx, err := s.pool.Begin(ctx)
 	if err != nil {
 		return WorkerAllocation{}, err
@@ -73,7 +78,7 @@ UPDATE gateway_worker_instances
 SET allocated_capacity = 0,
     updated_at = now()
 WHERE status <> 'active'
-   OR heartbeat_at <= now() - $1::interval`, workerHeartbeatStaleAfter.String()); err != nil {
+   OR heartbeat_at <= now() - $1::interval`, staleAfter.String()); err != nil {
 		return WorkerAllocation{}, err
 	}
 
@@ -83,7 +88,7 @@ FROM gateway_worker_instances
 WHERE status = 'active'
   AND heartbeat_at > now() - $1::interval
 ORDER BY instance_id ASC
-FOR UPDATE`, workerHeartbeatStaleAfter.String())
+FOR UPDATE`, staleAfter.String())
 	if err != nil {
 		return WorkerAllocation{}, err
 	}