The OCLC server claims that the xml payload is encoded as latin1 (ISO-8859-1).
This causes Requests to incorrectly encode things as latin1, when actually everything is (thank goodness) UTF-8.
We can fix it by just telling Requests that it is really UTF-8
With thanks to Tex Texin, creator of http://i18nqa.com/debug/utf8-debug.html
Together with #434 and #435, this fixes#429. Use the user-agent to
determine if the call is coming from a BookWyrm instance or not. If it's
not, give a pure activitypub representation for the status. Otherwise,
give a BookWyrm one, to allow for a complete integration between
instances.