Since hobbing using a tap gives a shallow worm-wheel effect, would it tend to try to push the belt sideways?
Might light, straight knurling be better, although possible only on a flat-belt pulley?
Perhaps the real answer is in the matching of power, belt, pulleys and belt tension. I’ve seen light knurling on the pulleys on small model engines, whose conditions of use are outside the designer’s control; but not elsewhere.
(A toothed-belt is a diffferent matter, used for definite angular relationships.)