vpc peering timeout due to subinterfaces not supported

[edipascale]

running the release test on env-3, in several peering tests the fabric state never converges due to the agent on the as4630-01 continuously restarting. Checking the agent logs, we can see that it fails repeatedly with:

time=2025-02-04T13:37:58.675Z level=DEBUG msg=Action idx=90 weight=33 summary="Create Subinterface Base 101" command=update path="/interfaces/interface[name=Ethernet16]/subinterfaces/subinterface[index=101]"
time=2025-02-04T13:37:58.796Z level=ERROR msg=Failed err="failed to run agent: failed to process agent config from k8s: failed to process agent config loaded from k8s: failed to apply actions: GNMI set request failed: gnmi set request failed: rpc error: code = InvalidArgument desc = SubInterfaces are not supported"

ideally, if some config is not supported on a device the pre-validation should detect it upfront, to avoid having to rely on timeouts and rollbacks.

[Frostman]

Yes, we need to validate against the features available in the switch profile. It's a bit tricky b/c the subinterfaces are used for the vpc loopback workaround