Allow Paperwork optionally to continue after potential Azure issues

With the environment/configuration variable OCR_CONTINUE_ON_ERROR set to true (default: false) Paperless will continue to process the document itself (i.e. OCR) if the Azure OCR failed for any reason, making the pipeline overall more robust (though you might miss the better OCR in such cases...).

Removal of non-transparent "safe-word"

Non-transparent "safe-word" in the beginning of the overlay (to force Paperless to recognize machine-readable text) actually not required.

Cleanup, removal of empty page handling

• Removed the empty page handling. It didn't actually contribute to the functionality and made the overall handling more error-prone
• Added a better traceback in case of unexpected errors
• Temporary files are created only within the tmp-folder that Paperless hands over
• Some code cleanup

Ignore non-PDF (for now...)

Ignores non-PDF files and lets Paperless handle those itself.

Logging path for docker setup

Sets a default for Docker setups if neither PAPERLESS_LOGGING_DIR nor PAPERLESS_DATA_DIR environment variables are set.

Ignore machine-readable PDFs

Skip Azure OCR if PDF is already machine-readable

Invisible overlay

Use a simple short „safe word“ in white color on to make PDFMiner recognize machine-readable text and so skip Paperless OCR. The actual complete OCR‘ed text will be rendered invisibly.

Initial Release

First release of Azure OCR pre_consume script for Paperless-ngx. Includes removal of empty pages and optionally content cutoff.