Revision 1349043

clustermanagerservice: make sure we actually get a type Instance back when doing getVmInfo
clustermanagerservice: don't accept notifications of "running" Exited machines
clustermanagerservice: clean syntax in places
clustermanagerservice: process state updates via checkInstances the same way as if the NM had told us (ie pick up on host and vmId changes).
clustermanagerservice: mark which functions are called externally
clustermanagerservice: make sure we copy the Instance instance for display in vmUpdate
nodemanagerservice: try to catch ValueError
nodemanagerservice: don't report source VM in migration exiting to the CM
nodemanagerservice: when the VMM notices a VM exit, it will tell the NM which will cause the removal of its record
nodemanagerservice: vmId is not yet known until the VM is received.