Convert Scanned PDFs into searchable PDF/A-3b formats

Create a Python script to convert a batch of scanned PDFs into searchable PDF/A-3b documents (containing an invisible text layer) using the open source doctTR OCR from Mindee. Prior to OCR itself, there needs to be a step for adaptive or dynamic PDF preprocessing to improve the quality of OCR. The notion of dynamic preprocessing means that for each document, the system first will evaluate what kind of preprocessing it needs based on the condition of the document. For example, if it needs de-skewing or rotation to correct the orientation, the system will automatically perform that; if the document needs noise removal or sharpening the system will automatically fix that. The output of docTR will be a PDF/A document which is then, in another step, finally converted to a searchable PDF/A-3b format (with an invisible text layer). This process is similar to what OCRmyPDF does except for the fact that OCRmyPDF uses Tesseract instead of docTR.