Gotcha. Thanks. I believe your car was also 370 rated (with 19" wheels) just prior to the website change that took it to 373 (and mine to 348), so I think that's the factor they were using at the time of your May snapshot. So it looks like Tesla does not use different constants for the EPA range/estimated mileage based on wheel selection.
Furthermore, I believe the SOC display when set to distance is based upon the full capacity including the buffer/reserve based on a couple datapoints I captured - the math doesn't seem to work if it was using the Usable capacity. When set to %, the SOC display excludes the buffer ([Usable - Buffer] / [Nominal Full - Buffer]). Also of note is that our cars have 5kWh of buffer whereas older ones always had 4kWh.
And to confirm, yes you're looking at the right thing (Nominal Full Pack) but just using the wrong units - it's kWh, not %. So your battery back in May had 95.1kWh, a little less than mine does (I'm at ~13k miles, about 10 months), and even less now. Looks like the "ideal" factory full capacity for our packs is around 98kWh.
If you frequently keep the car in a tight battery range (e.g. 50%-80%) then I would bet you don't actually have such significant degradation (which still isn't too bad), but rather the BMS has just drifted off on its estimate of capacity - there's no real way to tell how much capacity a battery has from a single measurement of voltage, so the BMS has to do coulomb counting and other magic to estimate the voltage vs capacity curve, which is non-linear to add further difficulty. Although it likely won't change anything in practice, you could drain down to 5% and do a full charge and probably increase your full pack estimate as the BMS recalculates how much energy it took to drain the voltage down to its lower threshold and fill it back up.