[evlatests] frequency offsets and phase

[Ken Sowinski]

Changes have been commited to deal with the observed phase change
with small frequency offsets.  For VLA antennas the approach used
is just what was done in Modcomp days: phase is calculated from
total delay = del_geom + del_cable.  For EVLA antennas, because
small changes are made by tuning an LO at the antenna, the LO
path length must be accounted for instead.  To do this the TDR
measurement on the fiber to each pad has been added to the database
and this is added to geomtrical delay to calculate the phase.  In
general this is close, but since this is a dabase parameter
adjustments can be made based on measurements.

The correctness of the phase calculation also depends on knowing
that heartbeat at the antenna is related to the heartbeat at the
control building by propagation time.  Unfortunately there is
some evidence that, at least form some antennas, this is not stable
at the level if a few microseconds, and sometimes much more.  Until
this is understood the utility of the corrected phase calculation
for EVLA antennas is in doubt.

The corrected phase calculation will be in use at the next software
update of the executor.