You are bending over backwards to come up with an excuse for Tesla, when they quite clearly made a choice to either be lazy or misleading.

When my gasoline car says "You have 12 miles left", I KNOW I can drive 12 miles minimum, as long as it isn't at wide open throttle. The car has mountains of data at it's disposal, but in reality this technique has been used ever since a car could figure out how much gas was flowing into the injectors. The assumption is that how you have driven for the past ten minutes is a reliable indicator for how you will drive over the next ten minutes, and for most ten minute periods that is an accurate assumption.

But the car can use your recent driving experience to get a good estimate. After all, if you've spent the past 10 minutes driving at 70 mph into a strong headwind, there's a decent chance that your next hour will be spent doing that. (Although this does get amusing on I-68 where my range estimates are markedly different after cresting the hill versus bottoming out in the valleys).

Sure, it's not the best estimate possible. But it's trivial to implement "assume recent past", and it's a far better estimate than "synthetic estimate that relies on a weighted average of two very different driving conditions."

Which concept more useful for drivers, the concept of remaining range if you keep driving the way you’ve been driving, or “how many miles the car can get in the EPA test cycle”?