Rosetta export fails because of graphical newline (u'\u21b5') character

[Christian Reis]

The rosetta export sometimes fails because a graphical newline (u'\u21b5') character is encountered in the translation content. This is a case similar to the dot character we use to indicate spaces; it needs to be transparently converted when saving the translation.

Of course, this only fails for pofiles in encodings that don't offer that character; even so, it should be fixed throughout.

  File "/srv/launchpad.net/production/launchpad/cronscripts/../lib/canonical/launchpad/components/poexport.py", line 67, in export_string
    chunks.extend([msgset.export_string() for msgset in msgsets])
  File "/srv/launchpad.net/production/launchpad/cronscripts/../lib/canonical/launchpad/components/poexport.py", line 155, in export_string
    return unicode(message).encode(self.pofile.header.charset)
UnicodeEncodeError: 'latin-1' codec can't encode character u'\u21b5' in position 171: ordinal not in range(256)