Some day might be interesting to track TSs for delivery to remote servers

If nothing else, just providing latency stats for talking to all the users in the room might be vaguely interesting

The issue names no file, test, or entry point; start by clarifying what timestamps should be tracked and whether the target is per-user or per-server latency. Before implementation, identify the relevant Matrix specification section and define measurable completion criteria for reporting delivery latency.

Written by the indexing model from the issue text.