I'm not sure you understand ... by your description that the belt is lifted by the flange -- even a small amount -- and then falls back down as it rotates around, that is called riding up the flange, and no timing belt should do that. No one is suggesting the belt is riding up on top of the flange. In a fixed center design, that isn't even possible unless those centers were reallllly far apart, the belt was extremely elastic, or there was grossly insufficient tension, none of which apply to these bikes.
And the image does not suggest the belt needs to move a mm to the right. I don't know where people get this idea the belt needs to be centered on either pulley, but it does not. I know Zero says something about 1 mm off the inner flange of the rear wheel. As someone who has to write instructions for equipment I design, I can tell you that number came out of someone's @ss because an instruction needed to be written and some number was needed to give people a target -- and flush against the flange is a poor instruction for reasons that are beyond this post, but you can likely figure it out. The position on the sprocket is irrelevant. The only way that number would be relevant would be if Zero did such a good job of aligning the motor and its sprocket in the frame relative to the rear wheel that 1mm off the rear flange coincided with the point of proper alignment. It's possible, but given some of the recalls I've seen, I doubt they went through that much trouble. The belt just needs to run true between the sprockets. If your belt is being lifted at all by that outer flange then yes, the alignment needs to be adjusted, but when properly adjusted, the belt might very well still track against that outer flange.
It's unfortunate your bike has always made this noise. Was it new when you got it?