Bugs in batch generation of fingerprints by rxnfp

[wangxr0526]

This job is really great, but I encountered some small problems when using rxnfp. https://github.com/rxn4chemistry/rxnfp/blob/master/rxnfp/transformer_fingerprints.py#L134 It seems that the last batch of data is discarded in generate_fingerprint when len(rxns)%batch_size!=0

This change may fix the problem:

def generate_fingerprints(rxns: List[str], fingerprint_generator:FingerprintGenerator, batch_size=1) -> np.array:
  fps = []
  n_batches = len(rxns) // batch_size
  if len(rxns) % batch_size != 0:
      n_batches += 1
  emb_iter = iter(rxns)
  for i in tqdm(range(n_batches)):
      batch = list(islice(emb_iter, batch_size))

      fps_batch = fingerprint_generator.convert_batch(batch)

      fps += fps_batch
  return np.array(fps)