Missing implementation of doc embeddings during inference

[valedica]

Document embeddings are not calculated during inference in neuralcoref.pyx, but they are left at zeros.

neuralcoref/neuralcoref/neuralcoref.pyx

Line 717 in 60338df

doc_embedding = numpy.zeros(SIZE_EMBEDDING, dtype='float32') # self.embeds.get_average_embedding(doc.c, 0, doc.length + 1, self.hashes.puncts)

This causes a mismatch between inference and training input features (doc embeddings during training are correctly calculated in document.py). Is it a bug or is it intentional? There is a call to a method get_average_embedding as a comment but it does not exist.