feat(运维监控): 增强监控功能和健康评分系统

后端改进：
- 新增健康评分计算服务（ops_health_score.go）
- 添加分布式锁支持（ops_advisory_lock.go）
- 优化指标采集和聚合逻辑
- 新增运维指标采集间隔配置（60-3600秒）
- 移除未使用的WebSocket查询token认证中间件
- 改进清理服务和告警评估逻辑

前端改进：
- 简化OpsDashboard组件结构
- 完善国际化文本（中英文）
- 新增运维监控相关API类型定义
- 添加运维指标采集间隔设置界面
- 优化错误详情模态框

测试：
- 添加健康评分单元测试
- 更新API契约测试

backend/internal/service/ops_aggregation_service.go

@@ -376,28 +376,37 @@ return 0
 `)
 
 func (s *OpsAggregationService) tryAcquireLeaderLock(ctx context.Context, key string, ttl time.Duration, logPrefix string) (func(), bool) {
-	if s == nil || s.redisClient == nil {
-		return nil, true
+	if s == nil {
+		return nil, false
 	}
 	if ctx == nil {
 		ctx = context.Background()
 	}
 
-	ok, err := s.redisClient.SetNX(ctx, key, s.instanceID, ttl).Result()
-	if err != nil {
-		// Fail-open: do not block single-instance deployments.
-		return nil, true
+	// Prefer Redis leader lock when available (multi-instance), but avoid stampeding
+	// the DB when Redis is flaky by falling back to a DB advisory lock.
+	if s.redisClient != nil {
+		ok, err := s.redisClient.SetNX(ctx, key, s.instanceID, ttl).Result()
+		if err == nil {
+			if !ok {
+				s.maybeLogSkip(logPrefix)
+				return nil, false
+			}
+			release := func() {
+				ctx2, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+				defer cancel()
+				_, _ = opsAggReleaseScript.Run(ctx2, s.redisClient, []string{key}, s.instanceID).Result()
+			}
+			return release, true
+		}
+		// Redis error: fall through to DB advisory lock.
 	}
+
+	release, ok := tryAcquireDBAdvisoryLock(ctx, s.db, hashAdvisoryLockID(key))
 	if !ok {
 		s.maybeLogSkip(logPrefix)
 		return nil, false
 	}
-
-	release := func() {
-		ctx2, cancel := context.WithTimeout(context.Background(), 2*time.Second)
-		defer cancel()
-		_, _ = opsAggReleaseScript.Run(ctx2, s.redisClient, []string{key}, s.instanceID).Result()
-	}
 	return release, true
 }