If they "firm up" it means they're outgassing; that's the only thing that could do that.
(FWIW, all pouch or prismatic cells require compression to correclty operate; the manufacturer spec sheet should call out what pressure that should be).
If the cells were truly "good" they'd be identical in characteristics. If RSD is Relative Standard Deviation, or what the worst case difference is between them, and 0.015% is the "average" difference between them, I guess that's pretty close to identical.
But the fact that the softer cells specifically are different from the others indicates a phyisical and electrical correlation--that there *is* a physical problem with them causing an electrical difference. That sort of thing usually gets worse with time.
I have decade-and-a-half-old EIG NMC cells that even now that their capacity and internal resistance has degraded significantly from aging, are still the *same* as each other--they still stay equal in voltage (within a flickering hundredth of a volt on a Fluke 77-IIIA) when charged or discharged, and I've never used a BMS or balancer on them, on any of the three 14s2p packs.
I don't have any test equipment for cell Ri, but the Cycle Analyst has an estimate of total pack resistance I could check to see what it presently thinks they are, and I've noted a few times over the years what it saw them as in the SB Cruiser thread, to see how they evolved over the years...but it's just an estimate, and only for the whole pack.