[evlatests] glitches and phase switching

[Ken Sowinski]

Due to wye-mon problems today only antenna 16 was available to 
point and that only after acu/mib problems were addressed.

I tried an experiment to learn something about the interaction 
between phase switching and the 'dropouts' Rick has reported.
The method was to turn off phase switching at the correlator 
for antenna 16 and see what comes through while looking at the
amplitude on a random baseline with 8 WG cycle integration time. 
I intended to do this with a Walsh pattern of alternating ones 
and zeros, but at first left the correct seed, 34, in the L302s.

1.  With a seed of 34 both IF pairs displayed an erroneously large
amplitude of about 1/4th the nominal amplitude every 1-2/3 seconds.
I set one of the L302s to use seed 1 and the response for that
IF pair changed to a glitch once every 10 seconds at the 10 second
tick.  Killing the executor stopped the glitch.  I first tried to
synchronize the antenna, just in case, but that made no difference.

2.  With the executor running again and the L302s using the correct
seed I set the walsh_offset to the nominal value -32/86400 = 0.000370...
That made the 1-2/3 second glitch go away independently for each 
IF pair.

There seem to be two problems.  
1.  The phase switching command to the DDS is wrong when a command
to the L302 is processed.
2.  Perhaps the the walsh_offset is off by a WG cycle.  Reading the
code suggests that it is set to -(32 + 1 WG cycle), but it today
-32 sedonds seemed to work better.

This should all be tried again with antenna 14 as well.