Yes, that helps. It is easy to see the mechanical gyroscope, along with the torques induced by earth rotation, and the associated precessional affects. I guess for the RLG, one simply takes the rotational component about each axis (determined by phase shifts, not mechanical torques), and sums them up to get the resultant vector, as you say. It is still impressive that there is enough sensitivity to detect the earths rotation, although I guess these devices are incredibly accurate in measuring phase shifts.
Thanks