the streaming trajectory relay actually drops traj pts when it loses connection. It claims to then "retry later", but it actually skips the point it dropped and continues with the next one in the trajectory. Before this change, that was not too much of a problem, as the connection would not come up again (requiring a restart, emptying queues and restarting everything).
While reviewing #263, there have been a few cases where the relay dropped points it claimed to "retry at a later time".
From https://github.com/ros-industrial/industrial_core/pull/263#issue-675682823:
this should be checked and fixed.