Genetic CFL: Optimization of Hyper-Parameters in Clustered Federated Learning