Refactor Cluster policies that trigger Kyverno restart on CRD installation

The following discussion from !4210 (merged) should be addressed:

• @tmmorin started a discussion:

  this mimics the 2 policies added by !1256 (diffs), which is very fine short-term, but not ideal longer term in terms of code-duplication

  can we agree to, as a follow-up to these two MRs, refactor these 3 policies into a single one (computing podAnnotations to include the name of the CRD having triggered the policy) ...?

Once the two MRs are merged, we need to refactor/re-write these specific policies, in order to avoid code duplication.