The metrics that we plan to gather naturally group into three types: metrics for all elections that the node has seen, metrics tracking the last election that the node was a candidate in, and metrics tracking the last election that the node voted in. We should use IDL types to be able to both easily update each of these metrics and easily report them in serverStatus or replSetGetStatus (without having to do any manual serialization). These IDL types should initially be skeletons, until we add more fields to track various metrics.