Just to let you know I tripped over a silly and somewhat exotic, yet
nasty rfsm_timeevents bug. It occurs when a timeevent does not result
in a state change (typically because of being inhibited by a
guard). In that case the timeevent will be raised for each step until
the state is exited (so possibly forever if nothing else happens).
The fixed version (to be found on the 'dev' branch) takes care that
this event is only fired once. This will be part of beta6. I hope
nobody relies on this broken behavior.