Design the state machine and RPC logic for a distributed consensus implementation based on the Raft algorithm. Specifically, detail the handling of leader election timeouts, log replication safety, and the cluster membership change process (single-server changes). Include edge cases for network partitions and split-brain scenarios.