Investigate high max latencies when running oltp workloads

[yhr]

Some high max latencies have been reported (~ 1500ms +) when running otlp workloads using percona server.

We do see high max latencies with xfs as well(in the same range), but it looks like for some benchmarks zenfs might be more likely to produce these outliers.

Let's try to figure out where the latencies are introduced by inspecting logs and adding instrumentation as a first step.