Octis

---------------
* Improve preprocessing 70
* Bug fix CTM num_topics 76
* Add top_words parameter to CTM model 84
* Add seed parameter to CTM 65
* Update some requirements
* Add testing for python 3.9 and remove 3.6
* Minor fixes

--------------------
* Update metadata Italian datasets
* Fix dataset encoding (57)
* Fix word embeddings topic coherence (58)
* Fix dataset name BBC_News (59)

--------------------
* Fix KL Divergence in diversity metrics (51, 52)

--------------------
* Bug fix optimizer evaluation with additional metrics (46)

--------------------
* Bug fix Coherence with word embeddings (43, 45)

--------------------
* ETM now supports different formats of word embeddings (36)
* Bug fix similarity measures (41)
* Minor fixes