wordcounter assumes stream is bytes

[huard]

Description

In the _handler, we have

        def words(f):
            for line in f:
                for word in wordre.findall(line.decode('UTF-8')):
                    yield word

which assumes line has a decode method, but the supported_format (TEXT) does not explicitly specify that the encoding is utf-8. So if the content is passed as an embedded string in the request, with no encoding information, the process fails.