Language-agnostic OCR

[pavel-denisov-fraunhofer]

I implemented a modification for the TesseractOcrModel to work with cases when the document language is not known in advance. It uses Tesseract's script detection to detect the script, and then runs an appropriate script OCR model (e.g. "Latin" for English or German). Would you be interested in this feature in Docling? If yes, I could prepare a PR.

[dolfim-ibm]

This is a good idea. I think we might have to review a bit the design for when to use the "auto" language mode, but it would definitely be a nice contribution.