My agent is processing PDF documents into Markdown using docling. Some of these PDF contain images and docling can use OCR to perform those actions. I have easyocr installed and docling configured to use it, however on each agent run the code calls out to docling and that+easyocr appear to be downlaoding the ML model to perform the OCR. Is there a way that I can have it download that on deployment so it is always available to each agent run? Does the deployment process have any post deployment script hooks that I can configure? Or some other mechanism?