Jira Link: DB-6564

Description

We should add some metrics around how many times we call it and how much memory we release at a time.
This is primarily useful for the tserver, but could also be valuable for PG, as we now use a MemTracker in PG and we've observed, in some benchmarks, that the PG time spent on the MemTracker can add up to 5% total time to the scan queries.
cc @spolitov @timothy-e

Warning: Please confirm that this issue does not contain any sensitive information

• I confirm this issue does not contain any sensitive information.