Optimize updates by moving Helm CRDs to runlevel 50

Cluster-version operator (CVO) manifests declaring a runlevel are supposed to use 0000_<runlevel>_<dash-separated-component>_<manifest_filename>, but since api#598 added 0000_10-helm-chart-repository.crd.yaml and api#1084 added 0000_10-project-helm-chart-repository.crd.yaml, those have diverged from that pattern, so the cluster-version operator will fail to parse their runlevel. They're still getting sorted into a runlevel around 10 by this code, but unless there are reasons that the CRD needs to be pushed early in an update, it gives the CVO the ability to parallelize the reconciliation with more sibling resources if you leave the runlevel prefix off in the API repository (after which these COPY lines would need a matching bump).