Extract structured JSON from OCRed document

[svenseeberg]

After the OCR step and classification, we should extract information adjusted for each document type. These JSON can then be sent to the front and, which in turn can ask additional questions to the user to improve answers.