If the interface is a miniport or filter, it is worthless anyway, and does not return any information of value. It is best practice to not monitor them as they just eat up polling resources.
What about the vmxnet3?
The vmxnet 3 - Ethernet Adapter - Prodnet and the vmxnet3 - Ethernet Adapter #2 - Backnet look like the actual NICs and would be the only ones I would keep.
Try switching the node from SNMP to WMI polling and see if the interfaces dedup better.
I've had lots of problems with discovery and SNMP on windows nodes. Everytime the node is rebooted and rebooted it rediscovers the interfaces again. That might be the source of your problem.
Are you listing resources after a reboot of the node? this maybe be a problem with the interfaces reindexing after the reboot? (possible Fix below)
you can delete the rogue interfaces by the following (but yea this will mean you will loose the interface history)
Click manage nodes.
Then re-list resources.
this could be your fix:
Is there a way to keep these indexes when the Windows server moves from one VM host to another?
My understanding is that each Virtual machine has its its own Universal unique identifier (UUID) this UUID is generated initially when the virtial machine is powered on. the virtual machine will get a new UUID if the VM is moved/copied to a new location (that could be why you are getting the unknown icon).
to configure your Virtual Machine to use the same UUID even if it is moved or copied
Simply power off the virtual machine
Open the virtual machine configuration (.vmx) file in a text editor.
Add the uuid.action property and set it to keep.
For example: uuid.action = "keep"
This information can also be found on the VMware Documentation Center
You can read up a slimier issues here (on a cisco device)