Crab, your argument sounds convincing, but to take the example of a teetering-head machine (ie no direct rolling moment applied to the mast by the rotor head - fuselage simply hangs pendulously), I would think the situation for a very high tail rotor would work like this:
1. Tail rotor drift still present as usual.
2. Left cyclic applied to cancel drift.
3. There is now a vertically-displaced couple between the rotor head side force on the mast (left) and the tail rotor side force (right). Tail rotor force is above rotor head, so the resultant fuselage roll is to the right! It will stop rolling when the C of G is displaced far enough laterally (leftward from under the mast head this time) to counterbalance the afore-mentioned couple. Obviously we now end up in the whacky situation where the fuselage has an angle of bank to the right, but the disc is angled to the left to stop the drift. Probably one good reason why they don't make them this way.
To put it another way, I'm saying that 'normal' mast head above T/R hub configuration will give you a left-skid low hover (nothwithstanding strange loading, crosswinds etc). If they are both at the same position vertically, there shouldn't be a roll. If the T/R hub was above the mast, I reckon it would roll right.
This is going purely on the relative positions of the forces applied to the system, and therefore the couples they form. The only forces are main rotor thrust, tail rotor thrust and weight, so to quote Marjie from the movie Fargo, 'I don't rightly agree with your police work, Vern!'
A bit of a bone of contention. Not saying I'm right ... perhaps I just can't see how I'm wrong, yet!