Requests recorded by TrafficRecorder include timestamps.

The initial replay client will not necessarily respect this timeline.

An initial "best effort" attempt to replay requests at the appropriate time point should be made.

This is not entirely accurate, given unpredictable network latency, thread scheduling, etc. but provides a baseline assessment to work from.

When mongor itself fails to submit a request in time, it should be able to report the delay between "expected time" (i.e., timestamp + (replay start - recording start)) and the actual time it submitted a request. This will quantify whether mongor itself is introducing sizeable delay as a result of its architecture.