I'm wondering what the standard procedure for helm charts involving CRDs, operators, and resources based on CRDs are.

For example, if CRDs are already applied, and a helm chart lists both a "deployment" for deploying an operator, as well as the actual objects/pods which use the CRD types - will both start at the same time? Are operators expected to eventually finish deploying then read the objects using the CRD? Will this cause timing issues?

Alternatively, is it expected for the operator to be deployed with a pre-install hook or job? And then the helm chart will deploy the dependent objects?

What is the standard thinking on this?