podman-remote: run --rm exits before container is truly removed

[edsantiago]

In window 1:

$ ./bin/podman system service --timeout=0

In window 2:

$ ./bin/podman-remote run --name myc alpine date;./bin/podman-remote commit myc myi;./bin/podman rm -a
Tue Jul 28 17:30:18 UTC 2020
3840e1d6deaadebc70f317f765e6ea015b67d222e575ff77f920f8f67adbaa23
480a2a915ba10c1eb8da260d8f721e07eb1093ba7baa125c86a04c3aab8a14f6

$ ./bin/podman-remote run --rm myi date;./bin/podman-remote rmi myi
Tue Jul 28 17:31:00 UTC 2020
Error: 1 error occurred:
        * could not remove image 3840e1d6deaadebc70f317f765e6ea015b67d222e575ff77f920f8f67adbaa23 as it is being used by 1 containers: image is being used

This is two sets of multiple commands on one line. It's not important to have the first set in one line, it just makes it easier to up-arrow to repeat, which you need to do because this is a race condition hence it doesn't always trigger. The second command, of course, must be done as a oneliner because the timing window is short.

A manual podman-remote rmi myi, a second or so after the above, works fine.

master @ 288ebec, tried only rootless because it would be a pain to add sudos to all the above and I see no point to bothering. I can test root if necessary.

[mheon]

We'll need to make the remote version of podman run call the Delete endpoint (with Force set to false) on the container once it exits iff --rm is set to true. If we get a "container is running" error, the user detached manually with the detach keys, and that's not an error.

[edsantiago]

I believe this was fixed by #7224 but the 'Fixes' line was in a comment, not OP