check for nil upgradeStrategy

We initialize node upgrade strateg in cluster store,
but on rancher upgrade if a cluster automatically gets
upgraded, it'll panic because of nil upgradeStrategy

No need to check for storageClass existence while creating a pvc

Force redeploy cluster-monitoring if workload doesn't exist

Revert "Refactor for lasso"

Refactor for lasso

Check app projectId during edit

[forwardport] Ignore error when adding indexer

Problem:
If a indexer already exists none of the controllers are registered so
clusterauthtokens do not get pushed downstream among other issues

Solution:
Ignore the indexer conflict error, this is expected after the first set
of controllers register.

Add introductory support for recognizing k3os clusters

When a k3s cluster is imported there is a possiblilty that it will be a
k3os cluster. In that case, k3s specific code may no longer be
applicable, ie: the cluster upgrade will require a different plan than a
normal k3s cluster.

Now, when a cluster is imported we will look for nodes with
"k3os.io/mode" labels on them to determine if the cluster is k3os.

Update helm version to 2.16.5

Helm version checks

Removed the pre and post K8s upgrade related env variables
Added addon verification related method calls as part of validate_k8s_service_images

Use Catalog TemplateLister to find out the latest version of system catalog apps

This double checks if manage agent is found and unmanaged agent can't be
cleaned up by managed agent due to race condition. Once managed agent
is found then the loop is immediately terminated.