Having owned and driven EV's for about 11 years,
I've only driven my Spark EV for three years before this. But I can say that the efficiency improves perhaps 5% in the summer, irrespective of any AC and heat use (I've often turned both off entirely to see what the projected range comes out to be - and the mi/kWh displayed improves in summer), and controlling for tire pressure (I keep them consistent throughout the year). What the exact numbers are I don't know, but I can say that the silly guess-o-meter in the Spark (which is based on prior driving and AC/heat use) always shows the highest numbers in the heat of the summer. It does provide mi/kWh as part of the trip meter, which also goes up in summer.
Higher temperatures reduce internal resistance. During the winter period, living in LA, our temperature was mostly (50-75F) which isn't much different than San Diego.
The most relevant factor is the nighttime temp, since the battery can be cold soaked to that (if you park in a garage it is less of an issue, but it may still also track the outside temp fairly well).
At Ontario, CA, in May, nighttime temps were in the mid-50s. Notably, last week (concurrent with the rollout!), the
nighttime low temperatures were in the high 70s! (July 25th minimum temp was 80 degrees - I remember that unpleasant evening, even in SD). Using your plot above, that is 13C to 26C. That takes internal resistance of the red line (not sure which line would be most representative) from ~120mOhms to ~70mOhms. It's not nothing. I have NO IDEA how much of an impact on efficiency that would have (I don't know how much of the loss is in the internal resistance of the battery), though.
One thing I've noted is that i'm not convinced the Trip meter counters track HVAC usage which helps understand the drive train efficiency more (than actual person usage). I've seen some inconsistencies among the various trip meters (especially one time I got stuck in a snow storm for 10 hours and drained my pack by 40% sitting in the snow).
The trip meter counts energy usage when you are not in Park, including from the HVAC. If you are sitting in Park, nothing is counted.
Anyway, we'll see, perhaps. I don't know how else to approach this problem and have some certainty but to look at fleet data (we can't switch back to the prior software build and control for it that way). The law of large numbers suggests that when looking at a bunch of different trips from a bunch of people nationwide (and worldwide perhaps), it'll be the best indicator of the effect of any change (not-withstanding any methodology changes on the app side). The differences due to some people driving in rain, others not, some pre-warming/pre-cooling, others not, some people driving fast, others not, all average out. The alternative is to have a well known benchmark course, but that's difficult too since winds and temperatures can change and they can make large differences.
I'd recommend checking your Wh/rmi constant and see whether it has changed, too. Hard to do accurately, unless you have a long continuous drive, though. I doubt it has changed, though - you'll get 230Wh/rmi.