Version 0.0.71 breaks ocr_only pdf recognition on some documents #485

Describe the bug
Starting with version 0.0.71 ocr_only pdf processing of some documents yields an empty result. Exactly the same code works fine with version 0.0.70.

To Reproduce

1. docker run --platform linux/x86_64 -p 8000:8000 -d --rm --name unstructured-api downloads.unstructured.io/unstructured-io/unstructured-api:0.0.71

curl -X 'POST' \
  'http://127.0.0.1:8000/general/v0/general' \
  -H 'accept: application/json' \
  -H 'Content-Type: multipart/form-data' \
  -F 'files=@4.pdf' \
  -F 'strategy=ocr_only' \
  -F 'languages=deu'

3. Now pull version 0.0.70 and try the same. In the first case the result is empty, in the second the result is correct and non-empty.

• Filetype: PDF
• Any additional API parameters: -

Environment:

• self-hosted API
• any client yields the same result

Additional context
Attaching the problematic PDF.
4.pdf