sane defaults for comparison about codespeed HOT 4 CLOSED

hexiom2 is way out of scale; it works on its own, and "everything else" works sort of ... still too squished side-to-side for me to be sure.

from codespeed.

hexiom2 and telco are kind of outliers. hexiom2 takes a long time on both versions, telco takes a long time on Python 2. Stefan Krah has already opened an issue for fixing telco, I'm not sure what we might do about hexiom2. In the meantime, deselecting both of them makes the rest a bit more comparable.

from codespeed.

Looking again, telco isn't nearly the outlier I'd thought it was (and I'm not sure where I picked up that impression).

I have gone through and removed hexiom2 and all of the python2-only benchmarks from the default view on the comparison page, which also mitigates #3.

from codespeed.

I think this is covered enough with the change in which benchmarks are shown by default.

from codespeed.