tests should not borg errors

[nothingmuch]

when debugging DanGould/pull/8, the actual failure was not readily apparent due to the error using assert!(response.status().is_success()).

a helper macro seems like it would help with this kind of thing? although in this case, when printed that error happened to be unclear (see #420, it was just "connection refused" but that was an internal error, the directory trying to connect to redis)