Record CPU, IO, Network, Lock, and application wait times at KV layer and rollup to SQL Statement layer

**Is your feature request related to a problem? Please describe.**

It is hard to predict which SQL statements are waiting or consuming resources. If we were to record the CPU time, IO wait times, ... at the KV layer and roll them up to `crdb_internal.statement_statistics` we would be able to better understand the workload characteristics.

Using this data, we could sample the statement statistics and observe statements over a time-series to show resource consumption over time.

Jira issue: CRDB-26476