add smiles canonicalisation, update tokens.txt

[sfluegel05]

This solves #117

• in the _read_data method of the ChemDataReader (which is used by all SMILES related datasets), read molecule with rdkit and output new, canonical SMILES
• canonicalisation can be disabled by setting reader_kwargs["canonicalize_smiles"] in the dataset to False.
• this PR also adds new tokens since many of the tokens in canonical SMILES don't appear in CHEBI (mostly because of changes like -- to-2)

note for #113: SMILES augmentation only works if canonicalisation is set to False

[aditya0by0]

note for #113: SMILES augmentation only works if canonicalisation is set to False

Sure will take this into consideration