Problem:  The BATMAN board would occasionally (and sometimes often)
power up without the VAD properly initialized, leading to Video RAM failures
which, by the nature of the test, would loop.  If these tests were bypassed 
by pushing buttons, the program would go on to hang (can't run without RAM)
and the watchdog would reinit everything, which would somehow clear the VAD 
problem.  

	Solution:  Modify the RAM tests so that if there is a failure, the
program goes back and reinits the VAD, just like after the watchdog timeout.
There is still somewhat of a scope loop for the RAM failure and the operator
can still bypass the test by pushing buttons.

	Observation:  If the VAD does not create VBLANK after power up, and
the processor gets lost before it can init the VAD into working, things hang
up.  Using an independent watchdog timer (ala 555) would eliminate having to
depend on the VAD to power up cleanly and permit the processor to reset the
VAD directly during init routines.