GZipMiddleware is incompatible with iterator-style HttpResponse objects

[James Henstridge]

An HttpResponse object that has been initialised with an iterator can only have its content read once. The first line of GZipMiddleware.process_response reads the response body and throws it away:

if response.status_code != 200 or len(response.content) < 200:
    return response

After this point, whether the middleware decides to compress the body or not, the content has been lost.

If the middleware encounters such a response, it should make sure it reads the body at most once.

[James Henstridge]

Sorry, looks like I filed a duplicate. I didn't missed #7581 in my searches.