Proposal: Add Speed Benchmarking to FAQ

[lockwo]

Effective benchmarking of diffrax speed has come up a lot (see #687, #592, #517, #179, #577, #94, #82, #76, etc.)

While they are not all the same, there are some commonalities to how speed testing should be done and how to get the most out of diffrax and such. I think the FAQ would benefit from some sort of "Why is diffrax slower than [X]?" section

[patrick-kidger]

This sounds like a good idea :D If nothing else we can just provide an FAQ link every time this comes up.

I'll write a draft and tag you on it?

[johannahaffner]

I second this!