ArgoCD fails to compare app if ServiceCatalog CRD is installed in target cluster

[alexmt]

Application controller fails to load list of servicecatalog CRD with following error:

unable to retrieve the complete list of server APIs: servicecatalog.k8s.io/v1beta1: the server is currently unable to handle the request

Looks like servicecatalog is different from regular CRD and api server might fail with timeout. We should investigate why timeout is happening and find way to exclude such CRDs from app reconciliation loop.

[jessesuen]

Only other place where I see this message happening is in:
kubernetes-retired/service-catalog#2104

In that instance, the service catalog pod did not even start up correctly. So I believe the root cause of this issue is environmental.

That said, we should still try to handle it gracefully.

[alexmt]

Agree, as a next step want to try to reproduce issue locally, see if we can reliably detect it (probably special error code etc) and gracefully handle it.

[jessesuen]

Fixed in #1018.

Duping to #1015.