Performance-wise, why is UNION typically more expensive than UNION ALL even when you know the two inputs share no rows?SELECT id FROM a UNION -- vs UNION ALL SELECT id FROM b;