one way would be to duplicate you phase detector but with a fixed 90 degree offset. My idea is based on if one signal is phase retarding so as to come into phase with the other, the output from your main detect would be 2.0, while the output from the second detector would be 1.4, now if the phase continues to retard the output from the main detector will fall, if the phase advances it will also fall. The output from the second detector would rise in one direction and fall in the other. So with an analoge window detector and a bit of logic you work out which direction the phase is changing and have an up/down counter to count up the number of times the phases change in the + direction and down count the number of times the phase changes in the - direction.
Frank