Data structure for molecules

I noticed some existing code bases from previous work. Since we are heading towards using `torch_geometric`, which itself has pretty complete data structures for graphs, should we just directly use those? Otherwise I suppose we need to write codes to port our structures to torch_geometric compatible ones.

Also naming is a bit not intuitive:
`neural_fp.py` is mostly on mol-graph data structures. I didn't see a ECFP (or other fingerprint) function.
`transformer.py` defines many torch nn modules, many of which can be found in `torch_geometric` I believe.

I suggest organizing things the same way as deepchem? Such as dividing into subfolders for data manipulation, fingerprint, nn models, etc.

Read neural_fp.py and transformer.py first, then compare their data structures and modules with torch_geometric and the organization used by DeepChem. The issue does not name tests or a specific entry point; scope is complete only when the maintainers agree on the structure, naming, and migration approach.

Written by the indexing model from the issue text.