Accept raw pdf bytes instead of just paths

Can you add support for taking in raw bytes from a pdf instead of just a path?

To handle multi-page documents, I am trying to submit each page individually. But writing to a temporary file seems unnecessary if you can just take in the bytes directly.

The issue identifies PDF paths as the current input, but names no file or test. Start by locating the API that accepts PDF paths and trace how the document is passed to Amazon Textract. Done means the same API accepts raw PDF bytes, including individual pages, without requiring a temporary file; add coverage beside the existing input tests.

Written by the indexing model from the issue text.