[evlatests] 3-bit resetting levels test

[Rick Perley]

    Due to a blunder of mine, the overnight 'definite' test of whether 
the new regimen for setting up the 3-bit system wasn't quite so definite.

    So we re-ran the test -- corrected -- at noon. 

    This test simply alternated back and forth between X-band and 
K-band, both using 3-bit mode.   The result below apply equally to both 
bands.  (These two bands are on opposite sides of the feed ring, so the 
test is as definite as we can make it). 

    1) For the initial gain setup,  the data (and the switched power) 
both settled down to stable values 8 seconds before the end of the 
1-minute LST duration initial scan. 

    2) For the subsequent 'requantizer' gain settings, the system 
stabilized 4 seconds before the end of the 20 second duration. 

    3) For the following 'science' scan, the new requantizer gains were 
applied within 1 second of the start of the scan. 

    Following these settings being applied, all PSum values were equal 
to 14 counts (+/- less than 1) for all antennas/subbands/polarizations 
that I looked at. 

    It all looks very good indeed.