[WIP] Output using hdmf-ml schema

[rly]

Within infer.py::in_memory_inference, output using the hdmf-ml structure ResultsTable instead of into a raw HDF5 file.

@ajtritt it looks like there is no longer code specific to handling the cases when args.classify == True or args.umap == True. Are these use cases no longer used?

[ajtritt]

@ajtritt it looks like there is no longer code specific to handling the cases when args.classify == True or args.umap == True. Are these use cases no longer used?

Correct, these are no longer used. Running UMAP (i.e args.umap) requires an n^2 operation, so needs to be handled more delicately, so I stopped doing it here. We're always doing classification now (as opposed regression), so it's no longer necessary.