This updates the getstakingreward and setstakingreward RPC so that an array of winners can be manually set. This makes it possible to allow 2 winners manually if there is a doubt about which is the correct one (e.g. if 2 nodes disagree).
Also use this to allow for more winners if we are missing connection to the selected proof, and leverage the remote proofs to determine if a proof looks abnormal.
Mining remains unchanged, only the node vote is affected.
Depends on D15685.