I've done some further analysis after a colleague has kindly supplied some historical data. I can combine that with my own database to figure out the energy density improvement trends vs cumulative cell production, which is a fairly standard way of analysing learning curves.
We can say that as an approximation the pack level automotive mainstream energy density in high volume commercial use is approximately half of the most advanced individual cell level density released for purchase.
So today the cutting edge cell is at ~700 Wh/kg (cell) and the leading mainstream use is ~ 300-350 Wh/kg (pack). Clearly one can pay extra and push up the price/performance curve at any given moment. The aircraft designers themselves make that point.
Looking ahead the technology improvement slope on energy density seems fairly linear trend. That is likely because power density is the priority mainstream performance metric that is being pursued, and (without looking at data) that is likely a power law trend. The current energy trend appears to be approx. 133 Wh/kg per 1 TWh cumulative increase. So the current 2.5TWh cumulative corresponds to the 700 Wh/kg peak cell or the 350 Wh/kg avge pack. Therefore the 10 TWh cumulative corresponds to 1700 Wh/kg peak cell, or 850 Wh/kg avge pack. The 2.5 TWh is 2023, and the 10 TWh is likely to be 2028.
So the proposed design is indeed sensible in the context of the 2030s timeline they have in mind. They have done this cell learning curve analysis themselves even if it is not explicitly written down in their paper.