Switch back to old dataset creation workflow

The new DRUF (Dataset Resource Upload First) workflow is an impediment to creating datasets. It forces the creation of a dummy resource in instances when all resources are going to be ETLed. Since it creates the dataset without letting the user set the dataset title/url/description/metadata, telling new publishers how to work around it adds complexity to the instructions.