Hi Jonathan,
HA or VMware high availability would only check for host failures. So if the host is not responing on the network, HA would kick in. With regards to the Virtual Machine network, as long as the host is pingable, the Virtual Machines would not be migrated off unless a DRS event has occurred (cumpute resources imbalance)
I know Beacon Probing can help on the management network, but even that will not help if the link goes down on the vSwitch 1 in your case
http://kb.vmware.com/kb/1005577
What you can do is essentially set up alarms for the vSwitch1 to fire off an event (email or alarm) if the link redundancy or failure occurs.
You could also set up a monitoring tool such as NAGIOS etc to check for Virtual Machine ping responses
Alart from that, I don't really think there's much you can do, for let's say you have a host with VM's on it and the VM's are connected to VM network on a seperate switch, even deleting the portgroup or vSwitch itself will not cause any issue to the working host
Regards
a