Lets us specify init flags

[CMajeri]

As discussed in issue #481, this PR makes the use of --init flags when using --docker-run more reliable by detecting the use.

Before landing

• Update docs/reference/changelog.md
• Link to applicable GH issues in the changelog

[ark3]

Your implementation (in container.py) is more complicated than I expected. Why modify parse_docker_args(...) at all? It seems all we need is an additional test in the if statement on (original) line 163. If docker_args does not already have an --init, then add one.

Sadly there is no straightforward approach to testing if you make the change that way... But that's a symptom of the poor organization of the code as a whole.

[CMajeri]

I thought about doing that until you mentioned the --publish. I figured it would make more sense to treat all "specfic" parameters in one place (and when/if that becomes too big, move them and change the way they're treated).

I'm okay with moving the if inside the run command, but that makes it untestable as you've said.
And I find it makes the code slightly muddier.

[CMajeri]

I refactored it as you asked, it's much simpler and smaller, but unfortunately it can't be tested that way.
I tried it out by printing the docker_command, and it seems to do what is expected.