I am trying to implement my own driver for the newly released Nvidia nemotron-parse and am finding a lot of page primitives to be unclear:
The request is, putting Pydantic descriptions with commonly-used data structures will go a long way in helping developers with Docling.