roachtest: use m2d.4xlarge for 8tb restore test

We see that on 2xlarge this test runs likely runs into its EBS bandwidth
limits. The easiest way to avoid that is to switch to a beefier machine,
which doubles its bandwidth limits.

We should also survive being bandwidth-limited, but currently don't do
reliably - this is tracked in cockroachdb#73376.

Epic: CRDB-25503
Release note: None

pkg/cmd/roachtest/tests/restore.go

@@ -532,7 +532,12 @@ func registerRestore(r registry.Registry) {
 		},
 		{
 			// The nightly 8TB Restore test.
-			hardware: makeHardwareSpecs(hardwareSpecs{nodes: 10, volumeSize: 2000}),
+			//
+			// NB: we use 16 CPUs to get better EBS bandwidth on AWS. With 8 CPUs, we get
+			// a c5d.xlarge at 287.50 MB/s base throughput, with 16 CPUs we get 593.75 MB/s.
+			//
+			// See: https://github.com/cockroachdb/cockroach/issues/97019#issuecomment-1452144587
+			hardware: makeHardwareSpecs(hardwareSpecs{cpus: 16, nodes: 10, volumeSize: 2000}),
 			backup: makeBackupSpecs(backupSpecs{
 				version:  "v22.2.1",
 				workload: tpceRestore{customers: 500000}}),