Fix killing of unresponsive containers by adding init process

Created by: mrnugget

Fixes https://github.com/sourcegraph/sourcegraph/issues/15314

This adds the --init flag to the docker run command so that the entrypoint process (bash in our case) is started by a init process. That fixes the behaviour of the interrupt signal simply having no effect.

From what I could gather that's because PID 1 (which is the --entrypoint in a Docker container) is protected and cannot be killed. But --init is given the PID1 is an init process, which forwards the signal to our entrypoint and kills it, exiting afterwards.

Resources:

• https://stackoverflow.com/questions/45718967/how-do-you-kill-a-docker-containers-default-command-without-killing-the-entire-c
• https://unix.stackexchange.com/questions/457649/unable-to-kill-process-with-pid-1-in-docker-container
• https://stackoverflow.com/questions/31538314/stopping-docker-container-from-inside