I clicked on your PDF link and got a Cloudflare captcha. I think that is the issue. In other words, the api can not download the document:
One option could be to white-list the OCR API endpoint IP address, so that no captcha is shown?