[evlatests] antenna 16 pointing

[Ken Sowinski]

> Before doing the sensitivity
> test and attempt was made to improve the pointing of antenna 16 by 
> accounting for the tilt error measured last week in the pad tilt 
> rather than the antenna tilt terms.  This did not seem to improve
> things but was left in place for the K band measurement.

This was still in place when a pointing run was done the night 
of Aug 25 at L band.  It was found that the tilt error we were 
trying to correct had doubled.  Clearly I don't understand the
signs yet.  I have now incorporated the change in pad tilt with
the opposite sign and modified the L band elevation collimation 
using the results of this pointing run.  

The August 22 X band pointing data fits still wanted to explain the
measurements with different tilts in Az and El.  This tendendency
was not so obvious in the L band data; perhaps because the poor
quality of L band poiting measurements makes it harder to see.

Over the top observations are still not correct.  When elevations
including OTT are allowed the pointing model fit becomes nonsense
with a large chi square and large post-fit residuals in both axes.  
The pre-fit residuals do not change significantly.