Dont use requote_path when dealing with redirects.

This was breaking redirect locations with a query string, by quoting the
question-mark.  Since the path will be passed through requote_path on
the next request anyway, doing it here seems unnecessary.

requests/models.py

@@ -214,7 +214,7 @@ class Request(object):
                 # Facilitate non-RFC2616-compliant 'location' headers
                 # (e.g. '/path/to/resource' instead of 'http://domain.tld/path/to/resource')
                 if not urlparse(url).netloc:
-                    url = urljoin(r.url, requote_path(url))
+                    url = urljoin(r.url, url)
 
                 # http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html#sec10.3.4
                 if r.status_code is codes.see_other: