We just discovered dataset names must be unique to an organization instead of a workspace. This is troublesome since we want to have and workspaces that run in parallel and have essentially the same setup, including datasets. But if the space has dataset then cannot and vice versa.

Even if we combined the and workspaces so they both have access to the dataset (which is non-ideal for other reasons), we'd run into a different issue: both the and upstream processes would update the same dataset potentially allowing the environment to corrupt the dataset before uses it.