Why? Should they do it for cruise control too?

Isn't it the case that the death rate per miles driven is lower in the Teslas than almost any other vehicle on the road?

No. This is Tesla spin. The Tesla is no safer than any other luxury vehicle which it shares it's demographics with.

They need to compare against cars in the same class, not the whole industry. Also, I’m not sure death per miles is the right metric. What matters is the accident rate while operating in cruise.

Also, it is an inherently screwed data collection, where people will drive manually in more dangerous situations - so you pretty much get a statistics on going on a straight line for hundreds of kilometers, except for the actually problematic intersection/people walking before after the car, etc.