the vm_inv table can basically be dumped ... all the *_INV tables are basically just a status get from vCenter ... if you recovered the table, the state may be absent for affected items.
I think you basically have to truncate all of the *_inv tables, followed by property_map ... then restart vCloud Director services. This will force a full inventory sweep and get the correct/current status.
If you want help in doing this, you can create a support request and the technical team can help out.