[evlatests] Modcomp-free fast switching: fake source gone.

[Vivek Dhawan]

Conditions: 1 hour total,  ~35min on source; X band.
             Fast switch (90 / 210 sec) on (calibrator / blank sky).
             MODCOMP free operation.

Objective: to see if there is data leaking between calibrator and
            the target at the scan edges. This is thought to be the
            cause of the fake source near field center, seen on
            EVLA-EVLA baselines when the modcomps were in control.

Result:    No fake source on any subset of baselines.
            A deeper integration should be done sometime, but it
            looks good so far.

            The previous solution (under modcomp rule) was to
            quack away the scan edges till the fake source faded.
            This was not necessary in the present case:
            The online flagging is overly aggressive, it flags
            about 30sec of data at the start of each scan, whereas
            only about 3-5 seconds actually deserve it. (I was
            going off source only 75" due to a typo in my file,
            I intended to go off much more.) Since the integration
            time was 1.6 sec, I could see the antennas settling
            with good time resolution. I saw no substantial
            difference between VLA/EVLA antennas.

            There were the usual woes of individual antennas/IFs
            not working, which i will skip since they are not of
            global interest.

Vivek.