Well, it seems that we are all correct on this one, which is nice
I eventually found this academic paper, which is quite interesting:
Here is a snip from its conclusion:
"This study presents the ageing results of 27 commercial graphite/LFP cells stored under 9
different conditions of temperature (30°C, 45°C, and 60°C) and nominal state of charge
(SOCnom 30%, 65%, and 100%). The extent of capacity loss was found to be directly related
to the storage temperature. It is most prominent for the most severe ageing condition (T = 60°C
and SOCnom = 100%). Storage SOC also influences the capacity loss, though it is of secondary
importance compared to storage temperature".
So, it would seem that the state of charge during storage over a period of months does affect the capacity, but only by a few percent. I'd guess that most activators don't store their cells for months at a time on a regular basis anyway. If you are planning to do that, it is probably good practice to draw off some of the charge, and keep it somewhere away from heat.