Import cannot recover after a failed import

[adborden]

The import job will detect collisions in the name and generate a random suffix so that it can retry creating the dataset. This likely creates a duplicate. I'm not sure that is desired. If the job is interrupted and you run it again, it doesn't pick up where it left off. Instead it creates duplicates for any datasets that were created successfully.