Why are multiple threads being inefficiently distributed in this example?

Question

I have a test why I can run to get about 5 seconds of 100% CPU usage time on one of my cores.

This is the script

from threading import Thread
from time