Currently, Model2Vec vectorization requires setting the method parameter to model2vec.

MinishLab/model2vec#134 added a model_type key to config.json. It should now be possible to inspect the config.json file when method isn't provided and infer that Model2Vec vectorization should be used.