Tatoeba data source should index by lemmas #15
Description
Tatoeba data source doesn't currently lemmatize tokens when it indexes them for example sentence search, but it should.
This can be seen in the _split_sentence function set in the __init__ method of TatoebaExampleSentences