Harmonize specutils with NDCube and other nddata-like objects that have a spectral axis

This is basically a statement of a goal from the April spectroscopic sprint week (which followed-on from previous discussions in other Astropy and Sunpy meet-ups): we definitely want to have a way to "view" `NDCube` and other nddata-compliant data types as `specutils` datatypes. There's an proof-of-concept experiment here: https://github.com/astrofrog/nddata-experiments that shows the current thinking of how this may best be done.

cc @nmearl @Cadair and @DanRyanIrish

Start by reading the linked nddata-experiments proof of concept and the issue's references to prior Astropy and SunPy discussions. Define what it means to view NDCube and other nddata-compliant types with a spectral axis as specutils datatypes; done requires an agreed approach and implementation scope.

Written by the indexing model from the issue text.