Understand how we will reference datasets in a post-rosie era

Picking up on comments by @pdearnshaw at the sprint 2 demo...

In the migration away from rosie for experiment management, we don't yet know how model development runs will be identified in MASS. Need to keep in the loop with GC and RAL as they develop a design for this.

And in the meantime, what about datasets that live in the `ens:` class in MASS. CDDS handles these, and would need to pass this to ESMValTool, most likely using variable label.

No files, tests, or entry points are identified. Start by reviewing the migration context and coordinating with GC and RAL on how model development runs will be identified in MASS; clarify how CDDS should pass datasets from the `ens:` class to ESMValTool and what a settled design considers done.

Written by the indexing model from the issue text.