Lost my original response to a browser crash so I'll try this again...
I have to say you've been doing some excellent troubleshooting, data collection and forum sleuthing. If you want more insight I would recommend Alldatadiy.com . This is what I use if I have problems and need guidance from a service manual, read instructions on replacing components, or inspecting connectors, electrical schematics, service bulletins, etc.
I wanted to say that of the 20/96 cells in your screenshots, three clear voltage groups are leading to a total cell variance (in this sample of 20 cells) of 650mV! #49 and #81 are the weakest in your sample of 20 cells, followed by the 3.2-3.3V group, followed by the majority in the 3.4-3.5V group.
As you were saying, the issue could be weak/inconsistent capacities in your pack, or it could be inconsistent self-discharge rates that have slowly widened the variance in voltages. What you are saying makes sense that a balance could fix the latter. The former would require replacements if you wanted to drive the car reliably below a certain SOC. However, besides a balance issue and a reset, stay open to the idea of a reprogramming of the HPCM2 and/or BMS and relearning of the cells and capacity.
The variance in an imbalanced pack might become a problem below 2 battery bars in the Spark, or 20-30% SOC, where the voltage in the discharge curve reaches its "knee" and rapidly collapses. I can back that up, here are two sources showing a chart of the no-load voltage of an NMC-type Li-ion battery with respect to its no-load state of charge. This knee is still present but far less exaggerated compared to typical Li-ion discharge curves under different rates. I chose these sources because you were sharing no-load voltage measurements.
Source 1: See page 4 fig. 5 of
"Fitting the OCV-SOC relationship of a battery lithium-ion using genetic algorithm method"
Source 2: See Fig 6b of "
Comparative Study of Equivalent Circuit Models Performance in Four Common Lithium-Ion Batteries: LFP, NMC, LMO, NCA"
From your cell data and this predictive data, we can see that cells in group 1 (#49,#81) are at 3-4% SOC, group 2 (#48,#80) physically next to them are at 11% and 16%, and the rest in group 3 are in the 24%-35% SOC range.
Regarding driving at low SOC, I wonder what the nature of P0AFA truly is. I don't know what SOC you encountered your issue at (well, we do) but the BMS allows individual cells to discharge down to a cut-off voltage of 2.5V. The driving behavior with the HPCM2 is to taper the max power down to 0kW to keep cells above this limit and still produce power for as long as possible so drivers aren't suddenly without power in a dangerous situation. My point is that my experience with very low cell voltages has never resulted in an "EV Not Ready" bricked state with error codes. I have driven to the cutoff and I was able to shut the car off, contemplate my next moves, turn it back on with "EV Ready" still green, and roll down the hill with regen and coast up to a level 2 public charger. I wonder if the BMS monitors anomalous cell balance and GM or the folks at the Volt forums chose to poorly word P0AFA "low batt voltage", or an out of balance or weak cell actually dropped far enough below the 2.5V low voltage cutoff and bricked your car until service could be done.
Please keep us updated! At your next convenience please capture the values of all 96 cells and the current car's SOC for a full picture, and periodically check if the voltages have self-discharged on their own over time.
