Primers pushed out past flush with rimless bottleneck case heads after firing; here's why.
When the firing pin smacks the primer, that force drives the case hard into the chamber shoulder; enough to usually set the case shoulder back a few thousandths. (Much like a full length sizing die does.)
Now the case head is further away from the bolt face. That space, whatever amount, is called head clearance. About .002" is ideal for long case life and accuracy on chambered rounds before firing.
As the primer fires, burns powder and pressure builds, the case walls expand and grip the chamber wall starting at the shoulder (thinnest part of the case) then working back. The pressure also pushes the primer out past the case head a few thousandths.
As pressure peaks and reaches maximum, the case stretches back until its head is against the bolt face. That pushes the primer back in flush with the case head. If there's enough pressure to do so.
The above happens with normal maximum loads. If the load is below maximum, it may not be enough to push the case head back against the bolt face. So the primer ends up sticking out past flush with the case head. I've seen .308 Win. loads about 8 to 10 percent below maximum start having their primers stay pushed out.
Good way to measure all of this is to first, prime some cases (no powder nor bullet) then measure their case headspace (head to shoulder reference). Pop them in the rifle and measure again. The difference is how much their shoulders get set back. How much is determined by the firing pin spring strength, head clearance (space between bolt face and case head when round's full forward in the chamber against its headspace point, primer cup hardness, how far the firing pin's allowed to stick out past the bolt face and the case shoulder angle, area and friction between case and chamber shoulder.
.30-06 cases have had as much as .007" shoulder set back. I've had Federal nickel plated .308 case shoulders set back that much, too. Sharper shoulder angles and more shoulder area reduce that setback. Belted and rimmed cases can also have this problem, but usually not any significant amount.
If the load produces lower than normal pressure, primers oft times will stick out past case heads. Full length sizing them correctly will often push their shoulders back forward enough so head clearance will still be small enough. If not, then incipient head separation will soon happen.
I think your loads that had primers backed out were too much below maximum to have enough pressure to seat them back into their pockets flush with the case head.
