agree. Today you can use this setup in theory already with ReFS block clonging. The problem is that the S2D/SOFS backend would not be optimal and you would have to balance it manually. SQL Cluster and Hyper-V itself are the only from Microsoft supported workloads, because they include code to balance the S2D/SOFS backend. Maybe this will change over time and Microsoft will support unstructured data and backup files directly as SOFS workloads. Until then you can run multiple VMs with Veeam Repository Extends on the S2D cluster and use our Scale Out Backup Repositories to make them available.
Other alternative to ma it HA is to create multiple Failover Cluster shares and again use them as extends in Veeam. If you Setup the Failover Cluster in a way that each S2D node hold one active role of the shares the load would be balanced accross them by Veeam. In case of a node die, the other member of the failover cluster will take over. (At this solution no SOFS is involved. Just File Server with Windows Failover Cluster on top of S2D.)