[python] python client returns a bad binary representation of string for text/plain responses

With this response type

    responses:
      200:
        description: success
        content:
           text/plain:
             schema:
               type: string
               example: ok

The you get the following response from the generated client:

"b'ok'"

This is due to the fact the retuned body is in bytes and then the deserialisation tries to make a string of it and in newer python this generates a repr of the bytes which corresponds to this string.

The proper way would be to decode the binary body according to charset returned or ascii for all content type "text/" before any fancy handling of the swagger specific types.

Then maybe also the str conversion in serialisation can be removed for string type.

I will send a pull request solving this in a moment.

Start with the generated Python client's response deserialization and serialization paths for text/plain responses. Reproduce the example with a string response, then verify that the client returns the decoded string rather than a bytes representation and that existing generated-client tests still pass.