[evlatests] ea03, 3-bit IF D

[Ken Sowinski]

Something seems to be broken with IF D in 3-bitmode; 8-bit looks
more normal.  The amplitude repsonse for this IF is reduced and
the state count histogram for both D1 and D2 is strongly biased
toward large positive values.  This suggests that the T304-D
output is not very gaussian.  The fact that both D1 and D2 show
the problem tends to put the problem upstream of the samplers.
This was seen at K band; other bands were not checked.