suggestion: nicer way of doing author magic

instead of retrieving the `'full_name'` field in `AuthorNameNormalizer` and `AuthorDistanceCalculator` one can just make the signature of the tokenization function to recieve a full record. Then, the callers can just put their own extractor for the tokenizer. Also update the docs with that.

If this seems, have a look in the contrib directory. It will make sense.
