New Task: Add chebi-20 dataset

[jackapbutler]

Overview

I will add the chebi-20 dataset from this paper which provides rows which map from "CID", "SMILES" and a natural language description of the particular molecule.

Basic template could be The molecule <CID> with smiles <SMILES> can be described as follows ____. This dataset is also already mentioned in the awesome-chemistry-datasets repository.

[jackapbutler]

In order to test out the functionality of Hugging Face datasets, I can try to upload this to the OpenBioML Hugging Face organisation and use that path within the transform.py script.