design blocks 42, completed 42, missing 0

=== reader gpt-6-astra: 42 blocks, 21 problems, infra-touched blocks 0, P1 failed (not correct or <1x) 5 | metric log1p
arm means (mean over problems of per-problem rep means): bare 0.954, show_all 0.886, bare2 1.238, attempt_then_consult 1.171, consult_on_failure 1.242, audit_then_consult 1.135, prior_then_consult 1.258
  PRIMARY attempt_then_consult - bare2             -0.067 [-0.148, +0.012]  problems up/down 5/14
  audit_then_consult - bare2                       -0.103 [-0.218, -0.008]  problems up/down 4/16
  consult_on_failure - bare2                       +0.004 [-0.034, +0.049]  problems up/down 2/2
  attempt_then_consult - prior_then_consult        -0.087 [-0.158, -0.024]  problems up/down 4/15
  audit_then_consult - attempt_then_consult        -0.036 [-0.099, +0.018]  problems up/down 9/11
  prior_then_consult - bare2                       +0.020 [-0.027, +0.069]  problems up/down 12/8
  show_all - bare                                  -0.067 [-0.176, +0.024]  problems up/down 9/12
  bare2 - bare                                     +0.284 [+0.167, +0.412]  problems up/down 17/1
  attempt_then_consult - bare                      +0.217 [+0.113, +0.334]  problems up/down 15/3
  audit_then_consult - bare                        +0.181 [+0.079, +0.295]  problems up/down 14/3
  consult_on_failure - bare                        +0.288 [+0.169, +0.419]  problems up/down 18/0
  prior_then_consult - bare                        +0.304 [+0.185, +0.434]  problems up/down 18/1
  attempt_then_consult - show_all                  +0.285 [+0.157, +0.430]  problems up/down 15/6
  consult_on_failure - prior_then_consult          -0.016 [-0.079, +0.047]  problems up/down 8/12
  audit_then_consult - prior_then_consult          -0.123 [-0.225, -0.037]  problems up/down 5/15
  diag P2_consult - P2_refine (raw second pass)    -0.031 [-0.140, +0.092]  n_problems 21
  diag P2_audit - P2_refine (raw second pass)      -0.069 [-0.206, +0.068]  n_problems 21
  diag P2_prior - P2_refine (raw second pass)      -0.019 [-0.111, +0.073]  n_problems 21
  diag S - P1 (raw pass evals)                     -0.073 [-0.181, +0.019]  n_problems 21
  selection (which pass each two-pass arm kept): bare2: {'P2_refine': 29, 'P1': 13}; attempt_then_consult: {'P1': 14, 'P2_consult': 28}; audit_then_consult: {'P2_audit': 26, 'P1': 16}; prior_then_consult: {'P1': 13, 'P2_prior': 29}