docs: align reference gate with comparison verdict

docs/reference-verifications/linear-static-3d-euler-beam-reference-verification.md

@@ -13,6 +13,7 @@
 - status: `pass-for-physics-evaluation`
 - owner_agent: `reference-verification-agent`
 - date: `2026-08-09`
+- review_fix_date: `2026-08-10`
 
 The prerequisite build/test report has status
 `pass-for-reference-verification`. This report applies only the approved Abaqus
@@ -127,11 +128,11 @@ maximum tolerance-normalized error, RMS error, norm error, and worst-row index w
 recalculated from the row ledger and matched exactly. Missing rows, extra rows,
 duplicate rows, nonfinite rows, and failed rows were all zero.
 
-The Step 26 sample acceptance probe reads `$metrics.status -eq 'pass'`, but the
-authoritative TASK-24 `ComparisonReport`, test contract, and serialized JSON define
-the verdict as the boolean `passed`. The generated JSON has no `status` field and
-has `passed=true`. This report uses `passed=true`; the stale sample-field mismatch
-does not change source, tests, or upstream contracts.
+The authoritative TASK-24 `ComparisonReport`, test contract, serialized JSON, and
+Step 26 acceptance probe all define the verdict as the boolean `passed`. Final
+review reproduced that the former `$metrics.status` sample failed against the valid
+artifact, corrected the phase probe to `$metrics.passed -eq $true`, and reran the
+exact acceptance chain successfully. The generated JSON has `passed=true`.
 
 ## Execution Evidence
 
@@ -278,17 +279,19 @@ model-coverage adequacy.
 - Abaqus_or_other_reference_solver_executed: `false`
 - owned_report_created: `true`
 - phase_index_step26_modified: `true`
+- phase_step26_acceptance_modified_for_review_fix: `true`
+- phase_index_step26_summary_modified_for_review_fix: `true`
 - generated_build_local_evidence: `true`, ignored under `.harness/build/`
 - notes: comparison and HDF5 inspection were read-only with respect to the
-  approved reference bundle; pre/post hashes, Git diff, and Git status agree
+  approved reference bundle; the final review fix changed only acceptance/report/index
+  documentation, and pre/post hashes, Git diff, and Git status agree
 
 ## Open Issues
 
-- Non-blocking acceptance-text mismatch: Step 26's sample PowerShell reads a
-  nonexistent string `status`; authoritative TASK-24 code/tests/JSON use boolean
-  `passed=true`. No upstream file was changed in this independent gate.
 - Non-blocking evidence typo: the Step 25 build/test report prints the input
   SHA-256 without the second `DB` after `...1DB`; the fresh 64-character pre/post
   value in this report matches the unchanged working-tree file. Reference Git
   diff/status and the end-to-end reference snapshot check are clean.
+- Resolved final-review finding: the Step 26 acceptance probe now reads the
+  authoritative boolean `passed`, and its exact command passes on the current artifact.
 - No issue blocks Physics Evaluation.