The JetStream, MotionMark, and Speedometer benchmarks are fairly "grounded in reality", don't you think? That's where the speed claims come from on this page: https://www.apple.com/safari/
I have no expertise in this, but I think it’s a safe bet that every major vendor can find benchmarks that favor their particular browser. (Edit: and https://news.ycombinator.com/item?id=25073625 should make it quite easy for Apple)
Not in this case, especially when Apple calls out Chrome and Firefox and they haven't said otherwise.
These are the industry standard benchmarks, so there's that.
Do note that browserbench.org is run by Apple. Which doesn't quite mean it's wrong but that's what WebKit specifically optimizes for.
Yeah, these are good indicators of real-world single thread performance. Way more useful than SPEC etc.