Air drag and possible eddy currents due conductivity and less than perfect symmetry are already said, but there's one more: The magnets are made of particles and are not 100% uniform. There repulsive force varies slightly during the rotation. At a moment the upper ring climbs and soon after that it falls. You may think that those microscopic rolls uphill and downhill compensate each other with no loss and the net effect to the torque is zero.
But it is not. The slight vibration causes slight electromagnetic radiation, because there's field changes which have non-zero 2nd degree time derivative (=accelerating changes). The effect is not like friction between solid surfaces, it's more like the mentioned air drag - depends on the rotation speed.
A physicist is needed to calculate the amount and the structure of the caused radiation in a practical system. Unfortunately I am not one. I guess the effect is not measurable when the rotation speed is any practically possible.