Hey Mike,
You're correct that during a vMotion operation there can be enough of a timeout to cause the cluster to detect a failure and move databases to the passive CCR node. The way to mitigate that is by adjusting the cluster heartbeat timeout settings to allow for a longer period during which the cluster will tolerate no heartbeats being received. You can find instructions on how to do that here (this article is for Exchange 2010 but the settings are Windows Failover Cluster settings and not Exchange specific):
The guidance from Microsoft regarding support may actually be true. It wasn't until 2011 that Microsoft officially started supporting vMotion and HA on Exchange 2010 and it's possible that they never actually went back and supported it for Exchange 2007. I don't know for certain but I suspect that you'd be unsupported if you used those vSphere features in your virtualized Exchange 2007 environment.
Yet another reason to move to Exchange 2010/2013! 
Matt