Fix false alarm for self-fork protection

[uprendis]

• Fix false positive detection of a parallel validator instance in a case of instant restart after a crash
• • A crash leads to a data loss due to non-flushed data on disk. On an instant restart, node may download its own recent events which may trigger parallel instance detection (recent external self-event -> emitted by a parallel instance). Now node doesn't trigger parallel instance detection for events which were created before node startup