Improve /api/v5/monitor_current and /api/v5/metrics performance.
Prior to this fix, these two APIs query clustered nodes in a loop. Now the queries are sent in parallel. The latency to return the query is dominated by the slowest node in the cluster.
Also added node parameter to /api/v5/monitor_current API to avoid cluster-wide query when only need data from one node.
For example, with ?aggregate=false&node=emqx@node1.domain.name, the API will return data only for one node.