fix(reliability): 采集器并发去重 + 备份恢复即时重载 + context 贯穿请求链

- CollectorService 安装/卸载按采集器互斥去重(重复触发不再并发 docker rm/run 互踩);删除后卸载失败留日志
- ApplyStagedBackup 替换文件后立即 Reload 内存态,消除「重启前旧内存落盘覆盖恢复文件」竞态;暂存文件名升级 lokistack.json(兼容旧名)
- GetUsage/Validate 全链路 context 贯穿:请求取消即中止 SSH/k8s 查询(异步调用点显式 Background)

backend/services/cluster.go

@@ -107,7 +107,7 @@ func (s *ClusterService) Create(name, description, configType, apiServer, kubeco
 	db.Unlock()
 	db.RequestSave()
 
-	go s.Validate(cluster.ID)
+	go s.Validate(context.Background(), cluster.ID)
 	return &cluster, nil
 }
 
@@ -222,12 +222,12 @@ func (s *ClusterService) Delete(id uint) error {
 	return utils.ErrNotFound
 }
 
-func (s *ClusterService) Validate(id uint) {
+func (s *ClusterService) Validate(ctx context.Context, id uint) {
 	cluster, err := s.Get(id)
 	if err != nil {
 		return
 	}
-	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	ctx, cancel := context.WithTimeout(ctx, 30*time.Second)
 	defer cancel()
 
 	info, err := GetClusterInfo(ctx, cluster)