Fix: Tokenizer is not able to encode triple bonds

Hi everyone,

i again found one minor bug in deepchems SMILES tokenizer. While tokenizing my dataset, i observed that the triple bond token ´#´ was not tokenized and instead simply leaved out by the SMILESTokenizer aswell as by the BasicSMILESTokenizer. I think this error occured because of the regex pattern, where a linebreak is placed directly in front of the ´#´. Removing the linebreak fixed it for me in a local copy of deepchem.