The only evidence I have is from the EPA documents showing the discharge event energy (I've posted that in the 2020 2019 2018 thread...) varies in the vehicles Tesla tests for the EPA. But even that can potentially be obfuscated by the vehicle shutting down 1-2kWh early during the test because it decides it has to shut down the battery (those tests run into the buffer and behavior can be unpredictable). As an example, the 2020 3P+ was run twice and had capacities that differed by 1.4kWh in the two separate tests (77.5kWh vs. 78.9kWh), likely due to buffer shutdown, not an actual capacity issue (I am assuming the same vehicle was used of course) - while the other (different) 3P vehicles tested had 79.5kWh and 78.4kWh.
As a more extreme specific example, the 2019 SR+ had 54.5kWh, and the 2020 SR+ had 52.6kWh. I think more typically the battery starts at 53kWh (including the buffer). I think this may be part of the reason they voluntarily derated the 2019 SR+ from 247 rated miles to 240 rated miles - it more closely reflected the "average" reality - 7rmi * 219Wh/rmi is nearly exactly 1.5kWh.
Thanks Alan. I'm assuming that by the 'buffer' you mean the undocumented and generally non-indexed capacity of all battery packs to discharge past an indicated zero? Curious what your understanding is of this buffer and how Tesla might view it also?