Manually get a list of ensemble members and forecast steps by looking at the GEFS docs

[JackKelly]

Encode these into explore_nwps/datasets/gefs.py? Or into explore_nwps/yaml/handwritten/gefs.yaml?
Maybe this handwritten yaml would be shared across hypergrib and explore_nwps.

[JackKelly]

On the question of where these should go...

For now, let's keep things simple. Let's not have separate handwritten and non-handwritten files. The main metadata will all be "handwritten" to start with, albeit with some copying-and-pasting!

As to where to keep this metadata... Maybe a repo dedicated just to metadata? One repo per NWP dataset? Or keep it in explore_nwps/metadata/gefs/ etc.? I think I'll go for the final option for now...