Thanks, Nick.
I can follow your line of reasoning, and it's consistent within itself, the thing in it that's got me stuffed is why the force produced by the main rotor doesn't act at the rotor head? I mean, it's a force, and it drags on the fuselage, and that's where it's attached.
Another thing that seems to throw a spanner in the works is the case you mentioned, where the tail rotor is level with the CofG.
If, as you say, the main thrust then passes through the CofG, then according to your view, the helicopter would be sitting wings level, in fixed wing speak, but with tail rotor thrust making it drift right.
Presumably we'd add some left cyclic to stop the drift, tilting the main thrust vector a bit right of the CofG, which would also create a left rolling moment. What then would stop that roll? If the body's rotating around the CofG, it wouldn't be weight shift, and as the tail rotor thrust passes through the CofG, it couldn't be that, either.
In my (possibly confused) world view, it's those varying couples coming to equilibrium that'd do it.