If we need to do bulk delete on millions of records in prod against spicedb, we need to make sure it does not cause the spicedb postgres database to lock or cause significant strain on our resources when other service providers are making check calls.

Ideally we would test this by inserting millions of host records into an ephemeral or perf setup, executing a bulk delete, and then simultaneously making check calls against non-host resources. This will help verify whether bulk delete is a prod safe operation and does not bring down authz.