Docling OCR is an optical character recognition package that uses LLM
models for text extraction from images. It provides a unified interface
for OCR operations within the Docling pipeline, supporting multiple OCR
backends and LLM-based text recognition approaches.