Comment 1 for bug 1567977

Presumably this is happening in part because the content type header is being ignored and the content is being treated as JSON and doesn't remotely begin to parse, spawning the 400 you'd expect then. However, there's an earlier stage of processing that should be checking the content-type.