Merge pull request #991 from mindprince/v100-2-4

GCE VMs with 2x and 4x V100 GPUs can now be created.

cluster-autoscaler/cloudprovider/gce/gpu.go

@@ -78,6 +78,8 @@ var (
 		},
 		"nvidia-tesla-v100": {
 			1: 12,
+			2: 24,
+			4: 48,
 			8: 96,
 		},
 	}