The MAO (Machine API Operator) e2e tests for multi-nic are not cleaning up machineset machines after test resulting in the node becoming ready during other machineset tests and causing test failures.

4.20 / 4.19

frequently.  

1. Review following test: https://prow.ci.openshift.org/view/gs/test-platform-results/logs/periodic-ci-openshift-release-master-nightly-4.20-e2e-vsphere-ovn-serial-runc/1938385603997470720     

e2e fails    

e2e passes    

In the output of the e2e, we can see:

`STEP: got 4 nodes, expecting 3 @ 06/27/25 02:08:20.756`

It seems that during this tests, an additional node became ready during the tests resulting in the initial node count to not match the replicas count of the worker machineset.  The timestamp of the machine that was created prior matched the timeframe where the multi-nic tests was running that scales up a machineset to validate machine config is correct.