[evlatests] Antenna wrap: (J)OBSERVE vs. the on-line system

[Michael Rupen]

I ran into an annoying problem this morning, to do with antenna wraps.

The first scan in a (stoptime) OBS file was a fast switching scan
between two sources, one at what OBS. reports as 272 degs. az., and the other 
at 265 degs. azimuth.

* (J)OBSERVE say the wrap is unambiguous -- setting ADR or ADL makes
   no difference.  I had ADL left over from a previous version and neglected
   to delete it in the scurry to get the file in to fill unused dynamic time.
   Didn't seem like it would matter.

* When executed, the observation tried to switch between what the
   operator reported as azimuths of 453 and 95 degs. [see below], i.e.,
   between on wrap and the other.  This was rightly considered a Bad Thing,
   and the operator basically tossed the calibrator out for the duration of
   that fast switching scan, to avoid wear & tear on the antennas (and I
   would add, to get _some_ data rather than none!).

   NOTE: the az. reported by the operator is 180 degs. off from what
     (J)OBSERVE uses; this is a known difference between antenna azimuth
     (what the operator sees) and geographical azimuth (what [J]OBSERVE
     reports).

So, why did (J)OBSERVE not catch this problem?

Enter the mighty Ken.  It turns out that the old (Modcomp) on-line system,
and (J)OBSERVE faithfully modeling that, had the limits hard-coded as
625 and 95 degrees (I confirmed this by trial-and-error in [J]OBSERVE).
By contrast, the current on-line system has the limits hard-coded as
630 and 90 degrees.  So in those magical 5-degree sections of the sky
(J)OBSERVE incorrectly predict the behavior of the VLA.

This is a pretty minor issue of course, but annoying; it would be worth
documenting inside JOBSERVE (maybe even fixing if that were as trivial as
it ought to be).

Of course one could also discuss whether the current behavior
of the on-line system is really to be preferred to what we did in the
past.  It's too late in the afternoon for me to have a useful opinion...

Cheers,

        Michael