The pully model is a good one. If you use a thin belt, the pullies rotate at speeds propotional to their diameters. If you use a thicker belt, do the pully speeds change? I don't think so.
I'd like to second your comment that this has made me think harder than I have in a while, and is dredging up all my old math and mechanical engineering which I sadly haven't exercised in a long time.
I've got a challenge now to show which of the two models is correct and more interestingly, why? I quite honestly can't show why your model of "the wheels diameter increases by the track thickness and behaves just like a bigger wheel" is wrong, if it even is, but I'll go draw some pictures and try to figure this out and convince myself, and hopefully some others, through math, how the heck this works.
Right now, I don't think the track simply creates bigger wheels. If it does, your math is correct and I agree with you. What I think is the correct model is that the track forms two concentric ovals (or approximately that shape). Like a race track, everyone knows that the inner edge of the track is a shorter distance around than the outer edge.
The drive wheels go around the inner surface of the track, and the progress around the inner surface is determined by the rotational speed of the wheels and their diameters. Because this progress around the inner surface is based on the drive wheel diameters, it is identical to the progress those same wheels would make on a flat surface, and the relative rotational speed of the front and back drive wheels would be the same as on a flat surface.
Now, lets look at what happens to the outer track surface. The outer surface is longer than the inner surface. When the drive wheels make one full rotation around the inner surface, one full rotation around the longer outer surface has also passed by. This means that the track driven vehicle will move faster for a given drive wheel speed since one drive wheel trip around the inner surface causes one full trip around the outer surface in contact with the ground.
All this said, I think the relative speeds of the drive wheels stay the same as on a flat surface because they are traveling around the inner surface of the track at a speed and rotational velocity determined by the drive wheel diameters and it is completely independent of the track thickness.
The traveling speed is faster based on the track thickness, and drive wheel torque is proportionally reduced as well.
I'm off to do the math to see if this holds together.