[evlatests] ea28 pointing improved

[Ken Sowinski]

On Mon, 24 Nov 2008, Doug Whiton wrote:

> We found that the bottom maypole was loose.  It has been tightened and
> three-balled.  Hopefully this was the problem.  The ant is ready to be
> tested.

I have previously reported that there were two obvious pointing
error signatures: an unmodelable dependence of az pointing on
elevation, and an unmodelable dependence of el pointing on
azimuth.  This fixed the first of the two problems.  The elevation
problem remains and can be seen in azimuth also, now that the
larger problem is repaired.

The remaining signature is that the measured elevation pointing
error takes on two states about 0.5 arc-min different depending
on elevation.  The transitions are around 90 degrees and 250 degrees
of azimuth.