We added an alarm called `OVNKubernetesNodeOVSOverflowKernel` in
https://issues.redhat.com/browse/OCPBUGS-10271.

This alarm is great and we need visibility into these packet drops. Actually, it's already surfacing some customer issues that would otherwise stay undetected.

The mild problem, however, is the naming. Technically, there are many possible reasons for the `ovs_vswitchd_dp_flows_lookup_lost` metric to increase, not just an overflow in the netlink socket (as the name of the alarm suggests). In fact, I have written a KB article listing some of them: https://access.redhat.com/articles/7115263.

I'm opening this bug for us to consider renaming it as something more accurate (and less scary), e.g: OVNKubernetesNodeOVSDpLostPacket.

Also, we could update the description to point this out.