Vorticey, absolutely right, hence when you have wings level and ball in the middle you are actually sliding a right a bit - this is inherent sideslip.
The only way to fly straight ie fuselage exactly pointing the way you are going is with a little left bank and the ball slightly out. Or you could have a piece of string (wool in the case of the Gazelle) mounted centrally on the outside of the cockpit which will indicate the true direction of the airflow (except in a climb where the whirl of the rotating downwash can alter it's accuracy). Or you could tilt the rotor mast in the opposite direction to TR thrust.
Back to the point - I hope we have sufficiently thrashed out that sideslip in one direction, almost always gives roll in the other direction (positive dihedral effect again) and that therefore, other than with a very high TR position, there is unlikely to be any rolling couple caused by the TR.
Nick, I guess the original question regarding roll doesn't have an answer - it just does it?