I realize it's dithering with a small detail, but "headspace" isn't "loaded cartridge length." The headspace could be 100% correct, but since it's assembled reloads, the overall length could simply be too much for that particular chamber.
In tapered cases like the 5.56, headspace is measured from the front of the taper/base of the neck to the base of the cartridge.
Overall length would be shoving the bullet into the rifling and causing it to engage, engraving marks into the jacket.
Dropping them in shows it's the force of the bolt slamming home that pushes the bullet into the rifling. If the headspace was too short, it wouldn't lock properly. If the headspace was too long, there wouldn't be extraction issues. The round would rattle in the chamber.
Loading rounds long is a common technique that is believed to improve accuracy. It has to do with minimizing the bullet getting cocked slightly off axis before it engages the rifling. The difficulty is that a self loading action can set back bullets that aren't crimped, and can push the nose slightly off axis if they aren't held firmly.
There are reloading jigs that measure the exactly how far off axis the bullet nose it, and then help adjust it to within .001" to improve group sizes. The idea is to have the bullet straight in the bore so it isn't predisposed to tip over when it jumps to the rifling.
It remains to be seen for the individual rifle which proves more accurate. Just because it's a generally accepted principle doesn't mean it works in that gun. Considering that long loads have problems feeding from short mags, the action cycling them into the chamber, and how much force is involved, it's a matter of whether a hand loaded manual action target gun procedure is relevant at all. It goes the military small arms ammo is crimped - to resist the battering.
Try the action with a commercially loaded round of ammo and see if the problem is still there. I expect it's the reloads.