Quite possibly ... and this is what's keeping my interest tuned on this thread.
Mostly agree. An expander ball/mandrel won't push _all_ of the donut to the outside of the case every time. Depending on your bullet seating depth, enough of the 'sprung back' interior donut could add some unwanted tension to the next bullet seated. Donuts tend to keep migrating towards the case mouth ... so, it's usually just a matter of time before they begin to affect one's reloads. And my issue with them is that they're not always a one-time problem. As long as you're observing brass flow ... that brass is flowing as you've described. The 'problem' is compounded by hardening and annealing ... and the understanding of those effects as you push them inside and pull them back with an expander ball/mandrel.
Stellar observations.
Definitely need to turn the OD of the neck and kiss the neck shoulder junction. Followed by
concentricity analysis, careful partial resizing, interior neck reaming (not even sure if this is affordable for this round ... if not, forget I ever made this reply ... lol), more analysis, full case sizing (including final neck sizing), seating a bullet, dykem from stem to stern, and chamber analysis from there (which requires all artificial tensioning effects being neutralized).
What I would do (if this problem was mine), is build a dummy round the way I've described above (as best I can). Feed a paperclip through the flash hole, bend the tip into a full circle ...
View attachment 537503
... straighten out the 'handle end" so it feeds down through the primer ejection hole in the press' ram, and _then_ seat the bullet. This contraption can be dropped into the chamber repeatedly so you can 'see' what is going on in there. You can push with the 'ball end' of the paperclip against the seated bullet and pull the round out of the chamber with ease.
Sounds like a lot ... but, if you're setup for it, everything described can be done in a single evening ... added benefit of finding the ideal starting point for your seating depth.
I never asked what the head stamp was on the OP's brass, but it is highly suspect.
If it's not Lapua ... if it's not Petersen ... (and even if it is) it has to be rolled across glass and eliminated from the pile if it wobbles (not sure if that's the right word here ... maybe staggers?). One can never (100%) correct for bad concentricity that's manufactured into an individual cartridge case. That'll run the entire length even though we are usually only looking at the case neck. The 'tell' is in using a ball micrometer measurement on your unfired brass. If you see inconsistencies as you turn and click, turn and click, turn and click ... you know you've got sucky brass. That determines whether or not you can effectively ream out the ID donut (which is ideal) or make repeated efforts to 'push' the donut to the OD and turn it off with a lathe (because one time ain't gonna be enough).
FWIW, when I buy a batch of brass and get it ready for my chamber ... I am always working incrementally with at least 20 pieces (sacrifice) ... 5 & 10 at a time. Really hurts when your batch is 'just a hundred' and sold at Lapua prices.