I didn't think stretching resulting in head separation was an issue with fireforming rimmed or belted cases, providing the headspace is correct. You definitely need to create a "defacto" headspace shoulder in rimless cases.

Normally the case will stretch because the firing pin will drive the entire case forward (minimal amount if the headspace is correct) and then the gas pressure holds the forward part of the case against the chamber and pushes the solid case head back against the boltface, stretching the brass just forward of the web (where the solid head becomes the start of the case wall ie the brass "thin" brass starts).

If the headspace is good on a belted or rimmed case, and the brass doesn't fill the chamber (ie as in fireforming from another cartridge) then the case will mainly blowout to fill the chamber. With tight headspace the case head will start and finish in the same place so there should be negligable stretching.

Rimless cases with small (ie weak) shoulders have to be checked for stretching, even with good headspace because the weak shoulder is set back by the force of the firing pin as it is forced into the front of the chamber - this effectively increases the headspace at the instant of firing. Two example cartridges that do this (from my limited experience) are the 8x57 and the 505 Gibbs. I have a couple of 8mms and the cases will either die from neck splits or incipent head separation, but I still get around 10 loads out of them. I know a guy with a 505Gibbs and, even though he has set his FLS die up for minimal resizing (to suit his rifle), he only gets 5 or 6 firings out of his cases and they all fail due to incipent head separation.

I never had head separation issues with strong shoulder cases eg 308, 30/06, 6.5x55, 270, 7x57.

I know some people that have had 303s with bad headspace and have used the shoulder for headspace, like in a rimless case, but the shoulder is still relatively weak and case life not good.