investigate and improve (if possible) network operator version upgrade duration

Related to the CI test case failure that cluster upgrade should be fast which has
a bug with more info, there seems to be a much longer time frame to upgrade
the network operator when using OVNKubernetes as opposed to OpenShift-SDN.

The test case in CI fails if the entire upgrade (all operators) take longer than
75m , which is sort of arbitrarily chosen, but it would be lazy to just keep
bumping that timeout and not investigate to find improvements or bugs that
can be fixed to keep upgrades from taking longer than they have to.