Tweak test weights for #slow issues

Problem/Motivation

#3630648: Improve balancing of CI test jobs improved test distribution between parallel jobs.

I think we might be able to weight the #slow jobs a bit differently.

Assuming average method times of 7s, a test with one 20s method doesn't need #slow (although it's annoying if it's the last test to run, but usually not a big deal), a test with ten 20s methods generally does, a test with one 120s method definitely does.

This means we tend to tag two kinds of tests as #slow - ones with lots of methods where the methods are say 50-100% slower than average, and ones with only a couple of methods where the methods are say 200-100% slower than average.

So, if we apply additional weight to #slow tests inverse to the number of methods, this might be more accurate, without having to add ##slow or ##slow (which was my previous idea which I didn't even want to open an issue for).

Steps to reproduce

Proposed resolution

Remaining tasks

User interface changes

Introduced terminology

API changes

Data model changes

Release notes snippet