Fix stale Checkov runner state between scans

[rnehoroshev]

Summary

• create fresh Checkov runner instances for each CheckovWhorf instance
• add regression coverage for consecutive Kubernetes scans in the same process
• align the upload_results override with Checkov's current signature

Problem

Checkov.__init__ shallow-copies a module-level list of instantiated runners. Some runners retain mutable scan state, so a later admission request can process manifest data left behind by an earlier request.

Solution

Replace the shallow-copied runners with newly constructed instances whenever Whorf creates a Checkov scanner.

[rnehoroshev]

@Saarett Could you review please? This is a critical fix.