When a review request e-mail had utf-8 in the body, it would fail to send

because of a UnicodeDecodeError while assembling the data that gets sent to the

server. The fix is a pretty trivial addition of some .encode() calls.
We hadn't caught this before because the test backend for e-mail in django

doesn't run through the same sort of internal processing as other backends do.

I've changed an existing test case and added a new one so that we use the

console e-mail backend while running the unicode tests, so that we exercise the

same machinery.

Ran unit tests. Before the encode changes, they failed. After, they succeed.