The current way we were handling upgrades of Helm releases could
catch race conditions because we checked for status and then ran
the upgrade.  It's possible that both upgrades pass the status
check and then upgrade at the same time.

Instead, it's probably more wise to just do the upgrade and skip
failing if we get any error that the task is already happening
or that the release already exists.

If there are any changes happening to node groups, we do not need
to deploy the Helm release for the auto scaler as there will not
be any changes to it.  This will reduce the possibility for any
race conditions happening with too many calls at once if there are
multiple node groups being created at once for example.