Run camp-full_map-nano-002/order

Run camp-full_map-nano-002/order: passed essentially everything (98%) -- this test did not challenge the model.

Of 80 scored trials (order query kind), the model passed 98%. That's saturation: the difficulty range probed here isn't finding this model's limits. Consider a harder preset (breaking_point or adversarial_belief) or a wider dimension range.

2 failures recorded; all 2 survived verification.

Notable failures

• record f4ff81b3 — asked "Which happened first: Dexter moving the cracked wrench to the kitchen or Felix's aside? Answer 'first' if the former, 'second' if the latter."
  expected: first
  actual: second

  the model recovered narration order, not the true event timestamps

• record 29c45b86 — asked "Which happened first: Iris moving the sturdy basket to the break room or Iris's aside? Answer 'first' if the former, 'second' if the latter."
  expected: first
  actual: second

  the model recovered narration order, not the true event timestamps