Jobs running inside a Docker are not appropriately terminated if the timeout goes off.

There is a message in job.log about the timeout going off, but it does nothing, and the job continues to run.

WARNING: Timeout exceeded [7200], send SIGTERM to [1859767] and give a chance for graceful termination
wait...

For example: in this job

It is 100% reproducable for all jobs with "run_in_docker" and "timeout" in the config. for example: build job config

To reproduce, the timeout could be set to a small value, and job started with:
python3 -m ci.praktika run "Build (amd_binary)"
or more light weight job:
python3 -m ci.praktika run "Style check"
The commands should work locally, only python3 and docker (with no sudo permissions) are required