You might be able to eliminate flyers by shooting the lightest 20% together as one batch and the heaviest 20% together as another batch. In theory, the ones at the lightest end could have more than one component that was lighter than typical, and similar for the heaviest at the other end. As others have mentioned, with multiple components, one component could be lighter and another heavier for a given cartridge, and you can't distinguish between those in a completed cartridge.
If you decide to do your experiment with a brick, I would split them into 5% groups by weight (25 cartridges in each group) for the lightest and heaviest 20% of the brick (lightest 5% in one group, next lightest 5% in another group, etc), and shoot 5 groups of 5 OR split them into groups of 10 if you want to shoot 10-shot groups. You probably need to shoot on a calm morning or indoors to be able to distinguish legitimate differences. I wouldn't be surprised if the groups for the extreme ends of cartridge weights were at least a little better than the average groups for that brick.
