`encoding.from_message` incorrectly parses multi-line encoded headers

[wolever]

>>> msg = email.message_from_string("…")
>>> msg['Subject']
'=?ISO-8859-1?B?U2VuZCBkYQ==?=\n =?ISO-8859-1?B?dGEgZnJvbS==?='
>>> email.header.decode_header(msg['Subject'])
[('Send data from', 'iso-8859-1')] # The correct header
>>> clean_msg = lamson.encoding.from_message(msg)
>>> clean_msg['Subject']
u'Send da'  # Uh oh! Subject is getting truncated