[evlatests] Data stability with subreflector rotation

[Rick Perley]

    The implementation of the 'subreflector rotation trick' revealed a 
bug which caused the first 10 or so seconds of unflagged data following 
a change of band to be bad.  Barry identified the origin of the problem, 
and a fix was installed with last week's update.
    I just took data today at Q and K bands (where the problem is most 
easy to detect) to see if indeed the unflagged data are of good 
quality.  They are.