SpaCy/Thinc Shim models

[koaning]

I would like to add spaCy models. Should be a great way to learn the internals.

• ScikitBOWSGD.v1
• ScikitBOWNaiveBayes.v1
• VowpalWabbit

I need to think about how I want to handle featurization for all of these models.

[koaning]

For scikit-learn, we can use the hashing trick to generate sparse features. These can then be used by Naive Bayes, PA Classifier or SGD Classifier. These all accept sparse features and have fit_partial implemented.