[REQ] Python file_type streaming upload

Is your feature request related to a problem? Please describe.

When using a large file, python's requests will serialise this to memory. If the form rendering is bigger than the available memory, this will fail.

Describe the solution you'd like

Analogue to https://pypi.org/project/requests-toolbelt/ a way where file like objects offered via a "open(..., 'rb')" or "open(..., 'rb', buffered=0)" interface are never encoded to a body prior to sending.

Describe alternatives you've considered

I was hoping it worked out of the box.

This is my guess that describes the actual problem, it materialises the file contents.

    def get_file_data_and_close_file(file_instance: io.IOBase) -> bytes:
        file_data = file_instance.read()
        file_instance.close()
        return file_data

Urllib3 upstream multipart streaming support.
https://github.com/urllib3/urllib3/pull/2331

Start with the described get_file_data_and_close_file function and inspect how generated Python clients build multipart request bodies. Read the linked urllib3 multipart streaming support pull request for the upstream constraints. Done means file-like uploads can stream without materialising their contents in memory.