Feature request: Custom attributes/pipeline for input to Language

[kfkelvinng]

The custom attributes/pipeline can be useful for following pipeline to obtain relevant context from:

• marking the speaker from input
• marking the title/headings from rich text document (HTML, etc)
• marking the text category given from dataset
• additional doc from previous parse
• additional pipeline within Language.make_doc for spell correction, merging tokens, etc

In the process of integrating neuralCoref (single speaker) to spaCy pipeline by redirecting the doc on this line in neuralcoref within my Pipe.__call__(doc).

In order to fully integrate neuralCoref (with multiple speakers) into spaCy the pipeline components, Language would need to be able to consume custom attributes for marking speaker to nlp(text). Something along the line of adding extension to text_holder._.speaker={...} before nlp(text_holder).

Since OntoNote5, the speaker is annotated. This could be useful for future CoNLL tasks.