Feature: add support for Nanonets-OCR-s

[luchatex]

The image recognition effect of Nanonets-OCR-s is extremely powerful. With docext, PDF image files can be recognized
https://github.com/NanoNets/docext
https://github.com/inferless/nanonets-ocr-s

[Goldziher]

hi there,

It looks interesting.

Feel free to update this issue into a full feature request.

Note - we support only permissive OSS software (no AGPL / GPV V3 stuff), and we want minimal dependencies - not large. so if the model itself can be incorporated with a small footprint, that would be the best approach.

Furthermore, i am not sure there is a point for a model for markdown OCR, this is counter intuitive. Markdown is plaintext and includes formatting inside it.