See First, Answer Later: Visual Evidence Pre-Alignment via Sufficiency-Driven RL

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.