Code review comment for ~ltrager/maas:lp1813019

The way sampledata works is it creates a number of nodes with 1-5 disks attached to each. Once the node is created it then applies a randomly selected storage_layout. We don't know which layout until the node and disks are created. So the only way we can raise the minimium disk size just for VMFS datastores is by trying it and when it fails recreating the disks with a larger size.