Improve the observability of CPU consumption in region reads

## Enhancement Task

Currently, we have recorded CPU usage at TiKV since the last region heartbeat and reported it to PD:

https://github.com/tikv/tikv/blob/95a06da9a9365ce5b32ec670eb6aac44ebb45229/components/raftstore-v2/src/worker/pd/region.rs#L141-L155

We can now get it directly through the region meta within PD:

https://github.com/tikv/pd/blob/c53f203465f59018bbf57eab1507bd49a3b8ef85/pkg/core/region.go#L692-L699

Currently, access is limited to APIs such as pd-ctl. To improve cluster diagnostics, we should enhance observability of CPU usage during region reads—such as by adding relevant metrics to Grafana—to better identify hotspots within the cluster.