@oldcrow/oldspook/corpone, I've been told before that I miss the forest for the trees, and I think you might have a little bit of that tendency too.....But I'm bored and awake so I'll bite (again).
Going with your geometry discussion....
In a perfect world, the "rolling" method SHOULD show any differences between the diameter of the two bearing surfaces of the pellet (greatest circumference of the head and the skirt). The theory being that by "rolling" a person could cull out the pellets that don't fall into the same ratio as the majority of their tin mates, or rather, cull out the edges of the bell curve of head/skirt diameter in a given tin of pellets.
A big factor that you're not accounting for, is the axis of the pellet. Recent posts from nolimits helped me envision this. Any random selection of flying pellets, ie spinning pellets, MUST have their axis points perfectly centered in those two bearing surfaces in order to fly predictably. ("fly predictably" equaling accuracy and precision)
These are exaggerated examples, but anything like below SHOULD be sortable through rolling (I'm picturing clown car wheels with the axle not in the center of the wheel). But you were able to make the case for how difficult it is to get repeatable results with the rolling process. So there are factors at play within the rolling process that make its application harder in practice than on paper (pun towards all your proofs and napkin math intended)
There's also the question of obturation that was raised in the previous couple of comments on this post.
These last illustrations show the types of changes from batch to batch that JSB is notorious for. They change skirt depths, waist diameters/skirt angles/skirt thicknesses/etc, while keeping the same weight and name/label. None of those changes are going to show up in a roll test if the diameters of the bearing surfaces stay the same as the previous batch (assuming the axis point is correctly located in the center of the pellet). In theory, the head and skirt sizes and ratios could remain the same, but the differences in the skirt thickness and center of gravity are going to change how the pellet obturates, and therefore, its flying shape.
They guys who really know what they're talking about when it comes to this stuff will tell you that sometimes a batch of pellets with no discernible defects will shoot like crap. I think you're at that crossroad between theory and application. The theories only get us so far, actual shooting and testing is still necessary to get concrete answers as to whether a particular batch of pellets will shoot well from a particular gun and a particular barrel at a particular speed and at a particular distance (or distances).
As bosco said, there are simply too many factors at play.
(edit: I just saw the "accuracy" comment you seem to have left me. Wow, drinking tonight? I've not left any accuracy comments, negative or otherwise, on any of your various profiles. And Mike's not my hero, respect the guy in the context of his achievements in the airgun community? Yes, but I don't know him well enough to categorize him as a "hero." For example, will he run into a burning building to save someone's life? Dunno.)
