The report is not detailed enough. For the BMW X7 there were only 3 fatalities. But it doesn't show the details.
The gov website query tool does not provide many details. But underlying raw data (FARS) has much more details than any one would imagine. It does provided more than any questions asked here. It includes fatality of Tesla drivers, passengers and many more details, if you ever read a police report, you have all of that in that one line of raw FARS data (except for name and dob). It is sobering to read though as each line has at least 1 fatality and very sad story involved.
As for model X and bmw X7, same is true for Lexus 570/600, these are very low volume cars, 3 or 4 fatality a year means a lot statistically per car exposure. That is one of reason IIHS use 3 year data to produce fatality by model/make report. Model 3 stat would be interesting.
What is puzzled for me is model y has US sales between 160k to 200k sales in 2021, and has 12 fatal accidents involved in 2021 (read: the 12 is not exact Tesla model y driver or passenger fatality, you have to look into raw data to get that number). While Lexus RX 350 has a confirmed sales of 115k for US 2021 sales and has 5 fatal accident involved in 2021. So per car exposure, Lexus RX appears as safe as model y, if not slightly safer. Same is true for Volvo XC90, bmw x5, mb gle or Audi Q7. So you get an overall picture.
With AP and FSD, I would expect Tesla model y fatality stat is solid and out of chart vs German luxury SUV, but I don't see that trend (only 1 year in 2021) so far from gov raw data.
Tesla's own safety quarterly report is not 3rd party verified and does not compare its own car vs other specific make/model. If I were a scientist, I would say I can not produce peer reviewed findings for their claim, understand their claim still holds some value.