Suggestion re introducing sc_group vs sc_dataset

[falexwolf]

Extending discussion here: #11

• This would change the sequence and first introduce sc_group & sc_dataset and then talk about the specification within sc_group.
• It would make identifications with existing formats on the level of sc_group and not on the level of the fields within sc_group.

There is also a question here regarding "allowing arbitrary collections of dataframes" vs. aligned multi-dimensional annotations (#3).

[ambrosejcarr]

These changes look good and I'd be happy to see them merged. Pending the nit below.

@aaronwolen Could you provide a second opinion? Is there anything here you think should be changed?

[aaronwolen]

Looks great. I agree that starting with descriptions of the higher-level components and including their anndata/seurat/bioc analogs will help readers get oriented more quickly.

Related but probably for a separate PR:

• I think we should bump-up the matrix.png diagram to come before Notes:
• and some of those bullets could be folded into the revised descriptions (e.g., There can be more than one X arrays...)

[ambrosejcarr]

Thanks for this suggestion @falexwolf !

[falexwolf]

Thanks for merging! 😄