I respectfully disagree, because if it never left stationary mode why was the GPS turned on, on geofence exit!?
Because we have additional trigger for motion - geofence.
Well, then don't you think that there should also be another trigger to becoming stationary if a device stays within a certain area for some time that it switches over to stationary mode!? Otherwise a device might switch to motion state based on geofence alone and never switches back to stationary just like in the reported bug? Which is a matter of time before happening on a device not if it happens
Maybe you have a better approach than what is suggested to resolve the bug. I just don't know why doesn't it always happen and what triggers it