Replaced Message.get_charset() with Message.get_content_charset():

Fixes the bug described here: repoze#41

Rationale:

get_content_charset() has been there in email.message.Message for a long
time now, returning the charset of the body of the Message (which is what we
are looking for there).

get_charset() has been deprecated for some time now, returns None by default
(even if a charset is set in the Content-Type header).

There are some other ways to retrieve the charset (get_params(),
get_param('charset'), message['Content-Type'].params) but get_content_charset()
is the only one that exists through all existing python versions, and it is
there even in the new email.message.EmailMessage class.

Author: WuShell

repoze/sendmail/encoding.py

@@ -71,7 +71,7 @@ def cleanup_message(message,
 
     payload = message.get_payload()
     if payload and isinstance(payload, text_type):
-        charset = message.get_charset()
+        charset = message.get_content_charset()
         if not charset:
             charset, encoded = best_charset(payload)
             message.set_payload(payload, charset=charset)