Describe somewhere what the benchmarks measure, and how to measure yourself

[RalfJung]

I just realized that "baseline" and "clean" are the exact opposite of what I thought they were. "clean" sounds like it starts with a clean (empty) cache, and "baseline" like it would be the baseline, minimal cost of an incremental run with a hot cache. But actually it's the other way around.

Could the timing pages (like this one) explain this somewhere? Like a note at the bottom, or a tooltip in the actual table, or so. Having space for such an explanation would anyway be useful, as that would also be a good space to describe how one can run that same benchmark against a local rustc -- which usually is the first thing I need to do when I see a regression.

(I am still working on figuring out how to reproduce this "incremental clean" regression. Took me at least an hour, at this point, due to running into issues like rust-lang/cargo#5918.)

[Mark-Simulacrum]

Fixed in b52f402.