Refactor filters for policies and resources

[viccuad]

The first iteration of audit-scanner is in a sense, merging several filtering stages together. While acceptable, we can iterate on its architecture.

Refactor internal/policies/fetcher.go & internal/resources/fetcher.go.

The following list is not exhaustive:

• Clarify the several MapReduce processes that happen: filtering and collecting, then auditing the resources with the specific policies. These happen for policies, namespaced resources, clusterwide resources, and policy results.
• Allow for easier mocking and unit-testing of each stage.
• Remove code duplication. Example, inside GetResourcesForPolicies() & GetClusterWideResourcesForPolicies() (and their tests).
• Obtain information on which policy makes a resource get added to the list. This will allow to print warnings on misconfigured policies with wrong spec.rules, and allows users to reconfigure those policies, or better know which CRDs they are missing.
• Allow for more parallelization.