FedGCN: Convergence-Communication Tradeoffs in Federated Training of Graph Convolutional Networks