Deskew without ocr – feature regression or doc problem

The help suggests --ocr-engine none for "image processing without text recognition":

  --ocr-engine {auto,tesseract,none}
                        OCR engine to use. 'auto' (default) selects the best
                        available engine. 'tesseract' uses Tesseract OCR.
                        'none' skips OCR entirely, useful for PDF/A conversion
                        or image processing without text recognition.

But together with a55ab05 that moved the deskew detection into tessaract, no deskewing takes place anymore when replacing --tesseract-timeout=0 with --ocr-engine none.

At least get_orientation() and get_deskew()from tessaracts.py should still be run with --ocr-engine none, or a warning added to when run – from a users perspective it makes no sense that the output of --clean-final changes with the selected ocr engine.

Decided to report separate from #650, but at least related to this request.