allow nova delete to continue if quantum port delete fails

while doing some testing, I noticed that when terminating a nova VM, if the quantum port delete failed because of a bug in quantum, nova would not delete the VM. It seemed bad to prevent the reclamation of compute resources because of an issue releasing network resources. Compounding the issue is that there's no tenant visible feedback for why the delete fails.

So instead I propose changing the v2 quantum integration code to work like (I believe) the v1 quantum integration worked, which is that if a port delete fails, we log it, but keep on deleting. This may leave orphaned ports in Quantum, but that seems ok, as it is quantum's fault if it could not properly delete a port.