Due to a race in podman 1.6.3, when containers are created in
parallel under high disk IO, their associated overlayFS mount point
might become invalid, and container-puppet would fail to run.
In order to avoid the bug, run all docker-puppet containers
sequentially to avoid concurrent container creation. Will still
pull the images concurrently, so the time loss due to lack of
parallelism is only limited to puppet runs.
Change-Id: I29a9ffc621bcbd5e1b09490c5fe09a363286bb8b
Closes-Bug: #1856324
acd176e811