Restore C0 verification context before replay after restart #1697

Audit reference: Recovery follow-up

What can go wrong

The C0 proof verifier kept committee slots and E3 BFV presets only in memory. After a snapshot restart, the lifecycle events that populated them could be outside replay, causing honest in-flight C0 proofs to be rejected.

When it happens

This occurs after CommitteeFinalized and CiphernodeSelected, when a node restarts before all C0 key-generation proofs are verified.

What changed

Finalized slots and the exact E3 parameter preset are restored from durable repositories before EventStore replay; malformed identities and unsupported parameters fail closed, and terminal cleanup remains active.

Why the change works

C0 verification receives the same request and committee context it had before restart without depending on already-snapshotted lifecycle events being replayed twice.

Checks

Focused restart tests intentionally omit those lifecycle events and prove an honest canonical C0 submission reaches normal verification; workspace tests and strict checks pass.