endpointslice code in ovn-k node causes high disruption times for routes during upgrades

As reported in https://issues.redhat.com/browse/TRT-468 and discussed on slack , high disruption times to console/oauth/registry during upgrade in aws were reported after we merged the code in downstream ovn-k where endpointslices were introduced for ovn-k node. 

The endpointslice code was reverted upstream and downstream to allow time for investigating this issue.
Upstream tracking issue: https://github.com/ovn-org/ovn-kubernetes/issues/3116