I have to say: at first this kind of aproaching the things, felt as a work around.
But after a while, you realise its not. Its the way to do it.
To avoid the most problems poeple ask about in the post ...
The general rule is this :
Let the behaviours work on a invisible object,
make your own sensors, and try to keep them on "bounding" or "point" sensing.
snap the face object to them.
This also has as general advantage ,
since (as you felt allready) the behaviours run out of sync with the events,
snapping objects to the behaviours WITH events ...
will close the gap between them, or it is the best possible attempt to do so.
cracks me up
oh and ty captain