Running doc2rag PaddleOCR-VL (static demo) π Extract and explore OCR text from healthβcheck documents