Remove the unused UpdateRecoveryStatus and ResendOutputTo recovery payloads

### Task Summary

Two of the three `RecoveryPayload` subtypes, `UpdateRecoveryStatus` and `ResendOutputTo`, have no sender and no handler. `NotifyFailedNode` — the only one still in use, sent from `AmberClient` on cluster-node failure — stays, as does the sealed trait.

**Provenance.** Both arrived with #1677 (2022-11-07, "Amber Fault Tolerance: Global Recovery and Detection") and were genuinely used: the coordinator matched on them with `case UpdateRecoveryStatus(isRecovering)` and `case ResendOutputTo(vid, ref)`. #2208 (2023-11-15, "Refactoring of amber engine") removed those handler arms along with the old recovery path; from that commit on, the only file mentioning either name is `RecoveryPayload.scala` itself. Dead for about three years.

They picked up unit tests in #4829 (2026-05-03) during the coverage work, which is why they look live.

Removing them also frees the `ActorRef` and `ActorVirtualIdentity` imports, which only `ResendOutputTo` used.

Pure deletion, no behaviour change: −39 lines.

### Task Type

- [x] Refactor / Cleanup