Avoid downloading unwanted OCR model files

### Question
Hello, i have a docker container that installs the latest docling version (2.96.0) aftert that in a later layer in my Dockerfile the models are downloaded:
`docling-tools models download --output-dir /workdir/models/docling`

However its been stuck at Downloading rapidocr torch chinese models... for a really long time, while I plan to use easyocr or tesseract and not rapidocr. Is there a way to only download docling with easyocr files or only docling files without having to download an extensive corpus of rapidocr?