In the past, bugs in the purge logic went unnoticed, causing outages. This PR adds a step to the E2E tests to verify that resources are actually purged after VMs are destroyed.