[evlatests] antenna 23 pointing

[Ken Sowinski]

There was a pointing run Friday night and antenna 23
was included.  The results for 23 are better than
the last time this was tried but there still seems
to be problems.  Nonetheless I think we can solve
for tilt well enough for our purposes, so I declare
my willingness to have the antenna moved whenever
everyone else agrees is a good time.

Some details.  The fit to the traditional pointing
model is still poor, which suggests that something
unmodeled is going on.  Adding the n*az terms for
n=2,3 results in large coefficients for those terms
but only a modest improvement in the fit.  Further
evidence that something unmodeled is going on.

I can think of four things to look at.
1.  a 3 dB tracking test
2.  be sure that the amplitude response is stable
     on the timescale of a pointing trial.  (this
     seems to be the case.)
3.  be sure that all four IFs are working in case the
     squint compensation in PEEK is not done right.
4.  try a long track on a single source of pointing
     trials to see that errors change smoothly.