Parallel `compose-down`

In some scenarios, it may be normal that `podman-compose down` will be unable to stop running containers with a `SIGTERM` and has to resort to a `SIGKILL`. However, when many containers are defined in the `podman-compose.yaml`, the stopping, waiting and killing is sequential per container, and takes **a lot** of time unnecessarily (`$timeout x #containers`).

I suggest to support running `podman stop` in parallel for all the containers.