Implement automatic cleanup when pull-through caching fails or its main distribution is removed

[lubosmj]

At the moment, when the pulling fails in the middle of the execution, we may end up having some left-overs in the database. Similarly, when a user decides to remove a pull-through caching distribution, we should give her a way to remove everything, including created sub-distributions, remotes, and repositories.

[lubosmj]

Once this block succeeds:

pulp_container/pulp_container/app/registry_api.py

Lines 313 to 334 in a832b9f

	with transaction.atomic():
	repository, _ = models.ContainerRepository.objects.get_or_create(
	name=path, retain_repo_versions=1
	)
	remote_data = _get_pull_through_remote_data(pull_through_cache_distribution)
	upstream_name = path.split(pull_through_cache_distribution.base_path, maxsplit=1)[1]
	remote, _ = models.ContainerRemote.objects.get_or_create(
	name=path,
	upstream_name=upstream_name.strip("/"),
	url=pull_through_cache_distribution.remote.url,
	**remote_data,
	)
	# TODO: Propagate the user's permissions and private flag from the pull-through
	# distribution to this distribution
	distribution, _ = models.ContainerDistribution.objects.get_or_create(
	name=path,
	base_path=path,
	remote=remote,
	repository=repository,
	)

we can end up having an orphaned repository-remote-distribution triplet if the actual remote repository is not reachable. Perhaps, we should first try to ping the remote repository to verify whether it is available or not.