2 sync frs's
cpus 1(m), 2
cctimer
1 minor @ 1000[ms]
1 process each

The action loop takes more than 1 second, so we get overruns,
which we recover from by one of three possible recovery methods:

-p) inject minor frame
-x) extend minor frame by some amount of time (10ms in this case)
-s) steal a bit of time from the next frame (10ms in this case)

The number of overruns is read and printed from the "action"
realtime routines. It's not the place we'd want to do this
from in a real realtime appl., but it provides a good stress
test.

We terminate after 20 loops.

