Optimize histogram metrics to reduce metrics data

[zimulala]

Enhancement

In TiDB clusters with numerous nodes and high request loads, the volume of metrics data can become substantial, particularly for Histogram metrics. We plan to adjust certain histograms that generate huge amounts of data.
Below are some methods to reduce metrics data by decreasing the number of buckets in histogram metrics:

• Adjust the bucket count of the following high-volume metrics to reduce metric data volume: tidb_tikvclient_request_seconds_bucket, tidb_tikvclient_cop_duration_seconds_bucket, and tidb_tikvclient_rpc_net_latency_seconds_bucket: metrics: adjust bucket count to reduce metrics data tikv/client-go#1609
• Increase the display unit of tidb_distsql_copr_resp_size_bucket to reduce the number of buckets: metrics: update distsql_copr_resp_size unit #59991