We need to formalize/aggregate the restoration flow for the SpiceDB pg datastore. The SOP should include instructions for bouncing pods to invalidate local caches (if needed) and using Kafka message replay to fix any data inconsistencies that occur during the recovery process.

DR Doc: https://docs.google.com/document/d/1IHXRdeBjtpd0j6VD-aJQHp6yTx0vmVBLZkEPfR50wM4/edit?tab=t.0#heading=h.kevnm59v6bck

Document the end-to-end “data restore from backup” flow (including a reference to the infrastructure-level restore procedures) in an SOP document.