Pxtextmining

Models now trained on subcategories (minor_categories). Narrowed focus to SVC and BERT models, development ceased on other model types.

Further refinements to API to utilise trained SVC model, with improved documentation. API also takes question_type (also referred to as additional_features in the pipeline), which improves performance on specific subcategories.

First multilabel release

Version deployed for demoing to partner trusts

This is the first "long-term" (pre)release, in that no further changes/additions to the package are anticipated in the near future.