[EM] Filling Unscheduled Vacancies With PR

[Bjarke Dahl Ebert]

> With STV the results could be recalculated with preferences for the
> original winner ignored. There's a chance that could upset the rest of
> the seats though (non-monotonicity).

How about this: Recount using STV, ignoring the vacating candidate, but with one
difference: Never eliminate any candidate that was originally elected in the old
count.

I assume that the STV count algorithm will still eventually terminate (instead of
hanging in a state where no candidate can be eliminated), but that should be easy to
prove, or what?

-Bjarke