It is possible for a moveChunk command to receive a CallbackCanceled error response when the node is shutting down. We should have the network_error_and_txn_override.js override retry running the moveChunk command on a CallbackCanceled error response so it can learn the definitive result of the operation, similar to what we do with the shardCollection command already.