Although I wasn't able to support ignoring stale ballots, I was able to parameterize some behavior in the library -- this PR permits us to either error out or assume unknown candidates are in last place.
Coverage increased (+0.2%) to 86.538% when pulling 6a2aadc3ec129168b542fb9023e6889d801c2ab6 on carlschroedl:parameterizeUnknownCandidateBehavior into c44aa104fb49077238b3afe88654ec0a0714ab19 on pivot-libre:0.x.
Although I wasn't able to support ignoring stale ballots, I was able to parameterize some behavior in the library -- this PR permits us to either error out or assume unknown candidates are in last place.