Hi!
Well, now you got me going..
I can tell you we have never tried this setup with more than two hyper-v cluster nodes so far.
I can tell you that cluster communication needs to be disallowed on iscsi networks, switches need to be connected to each other.
The iscsi initiator needs to be connected to one virtual ip with default settings, then when connecting the target use multipath option, use microsoft iscsi initiator as adapter, correct source and destination ips (as in the portal).
then you add session to already connected target using again microsoft iscsi adapter, the second initiator ip, and second destination ip.
This assumes you have two virtual ips configured on SANs.
Thats for the front end config, off the top of my head.

I think you are right on the money there with the cause - the reservations not replicating can cause BSOD or reset on VMs - all depending on what they were doing at the failover time. In the past two and a half years of using this we have had plenty of strange things happen, but have usually managed to pin them to some other part of the setup. We have also been updating the dss v6 fairly regularly. I, for the life of me, can't say what thing/action/upgrade stabilized this for us...
In any case, I too hope open-e can resolve this rather critical problem with a patch soon. I won't switch to VMWare anytime soon - but I might lose the SANs given what 2012 is capable of.