Proactively delete old Docker images and dangling containers

[pietroalbini]

Over time the filesystem of a Crater agent fills up with previous Docker images and containers that failed to be cleaned up. Purging those on crater-azure-1 today freed up around 120GB of storage. We should have a cron somewhere (either on the agent code or on a proper cron in the server) that does the equivalent of:

docker ps -a | grep Exited | awk '{print($1)}' | xargs docker rm
docker image prune

For context, containers that failed to be cleaned up exist because of how Rustwide starts containers: docker create + docker start + docker rm. Rustwide is supposed to clean those containers up even if the Rust code panics or the build failed, but apparently sometimes docker rm itself fails.