After some helpful pointers from Christian Wehner I've started building a custom workflow to provision VMs into SDRS clusters.
HOwever, I can already see that this will mean a LOT of extra work. The storagePod object works fundamentally different from the datastore object, and requires a rewrite of the entire chain from the cloneVM action through every workflow/action between it and the topmost invocation.
I realize that priorities need to be done but I think the SDRS cluster object is the only Correct Way to handle VMFS going ahead, and anyone big enough to orchestrate VM creation will almost certainly also either already be using, or plan to use, SDRS to eliminate manual work from the VM placement process.
So with that in mind, I hereby humbly request that some time is set aside to rewrite most actions and workflows related to VM provisioning to support both DataStore and storagePod objects, preferably in a fashion which allows you to supply either (eg if ( storagePod != null) etc)
