Add a Macro Benchmark that Exercises Threading (Multi-Core Concurrency)

From @methane (https://mail.python.org/archives/list/python-dev@python.org/message/KVH7QLJ4VJBQQ45LZTWDXO2SEC6ANX7T/):

CPU cache invalidation exists regardless. With the current GIL the
effect it is reduced significantly.

It's an interesting point. We can not see the benefit from pypeformance, because it doesn't use much data and it runs one process
at a time. So the pyperformance can not make enough stress to the last level cache which is shared by many cores.

We need multiprocess performance benchmark apart from pyperformance, to stress the last level cache from multiple cores. It helps not only this PEP, but also optimizing containers like dict and set.

Commencez par examiner la suite de benchmarks pyperformance existante et la manière dont ses benchmarks à processus unique sont organisés. Définissez comment un benchmark de multiprocessing doit exercer une pression sur le cache de dernier niveau partagé entre les cœurs, et considérez le benchmark comme terminé lorsqu’il fournit des données de performances multicœurs reproductibles et utiles pour évaluer les optimisations de dict et set.