Restore the full logical clock from durable replay history #1669

Audit reference: I25

What can go wrong

After restart, the hybrid logical clock could resume from an old snapshot and assign timestamps earlier than events already durable in the post-snapshot log.

When it happens

The gap occurs when a snapshot is stale but later events are replayed before the node returns to live mode.

What changed

Clock recovery now considers both snapshot state and the maximum complete packed timestamp in replayed history.

Why the change works

The first new event is ordered after every durable event, including logical-counter ties within the same physical time.

Checks

Physical/logical time, overflow, restart, and stale-snapshot backlog tests pass.