Summary:
Our operator monitoring code would set the updating phase under certain conditions which might actually be failures. When the crd lists as updating,
the vizier_controller chooses not to run an update which can get the operator stuck. We basically should not tell the controller that we're updating in the
monitoring code because they should be separate.

In a future diff we will add a new state that tracks the reconciler status - that state will contain the updating phase (ready, updating, failed)
which will then be used by the reconciler to make decisions about whether to rerun updates or not. For now we just change this to simplify an otherwise
annoying loop.

Test Plan: Updated tests

Reviewers: vihang, michelle

Reviewed By: vihang, michelle

Differential Revision: https://phab.corp.pixielabs.ai/D9771

GitOrigin-RevId: 248d68e7d7abc7cbb917dc2f0f39e9a33e144a33