I disagree. The O.D. of the race is smaller than the hub. You must freeze the race if you want it to fit at room temperature. It will expand and effectively be clamped in place by the hub. When a mechanic "hammers" the race into place with a drift, he is attempting to overcome the interference fit between the race and the hub. There is quite a bit of force that secures this race and there is zero clearance between the OD of the race and the hub.

Tightening down on the adjusting nut won't move the race within the hub whatsoever.

The procedure of tightening the adjusting nut past the desired operational point is to ensure that the hub is fully seated against the inner bearing. If you simply brought the nut up against the face of the outer bearing, you could be misled by any clearance that existed between the rotor/hub and the inner bearing. Once all the clearance is definitely zero (or negative), then you can back the nut off and set it properly to the desired operational point (.0005" end play IIRC).

You could use a dual set of angular contact ball bearings like I do when I manufacture my spindles for balancing jet engines. The set costs $3000. But, you just clamp the outer races together and the two bearings load themselves up with the proper preload and the internal clearance is zero. The bearings run fantastic in this situation and they are eminently suitable for measuring unbalance that approaches .000010" displacement of C.G. Truly amazing. This could never be done with a tapered roller bearing that requires clearance to run safely.