[evlatests] Round-trip phase.

[Vivek Dhawan]

March 15-27: Delay comparison was made between 2 paths,
out on 1 fiber and back on 2, all in same bundle, ant.13.
Each path was measured by 2 meters. The 2 paths differed
by 10-20ps (out of a daily variation of ~100ps). The 2
meters on a given path agreed to within 1ps. This condition
caused concern, as half the round-trip phase could not be
used to correct for the 1-way phase if 2 fibers differed
by over 1ps.

Mar 28: starting march 28, the setup was moved to ant. 16.
the total delay variation was somewhat less, but the
divergence between paths remained.

Mar29-apr06: Archive was down, no data recoverable.

Apr6-9: archive back up. No one admits to altering the
setup. The EVLA's resident elves must therefore be credited
with the apparent improvement: The 2 paths now differ by
less than 1ps.  In addition, all 4 meters have the same
sign (since abuot march 15), I believe that was done
intentionally but I'm not sure by whom.

Further tests to be discussed in today's meeting.

Vivek.