[evlatests] Baseline 23 x 25 Delay

[Rick Perley]

    I'm perusing the (good) X-band data very carefully. 

    Baseline 23 x 25 has an extraordinary characteristic:  After all 
calibration is completed (and with very satisfactory results), this 
baseline alone shows a phase slope across all four sub-bands.  The phase 
slope is *exactly* two turns across the 128 MHz.  This corresponds to a 
delay of 15.6375 nsec -- and it is only seen on this one baseline. 

    I recall having reported on this before, on a different database.