fix(ci): refresh lab harness hash after e2e update

Record the workflow-pin change in the lab's policy manifest and keep the Unreleased changelog headings lint-clean.

Plan: IEP-EVAL-EVOLUTION-001\nBead: bd_000-projects-htjt.10.1\nRepo: intent-eval-lab\nEvidence: scripts/audit-harness init; scripts/audit-harness verify; markdownlint-cli2 0.22.1; prettier 3.8.4; e2e 21 passed

Author: jeremylongshore

.harness-hash

@@ -8,7 +8,7 @@ d29a2e9b94f7fd94f4f8953314323f96ca424cbaa6a985f6783b48ee2feae0f8  .audit-harness
 5b912f59eeef3c9b5f7de98d8ae27a0d8b2eeb03411f34d9fa0e205d061b7f7f  .github/workflows/ci.yml
 a349d30f7da0d4f6d2b6c7b6011115b8c0e474a30f820bc1354a7efaa10b2c37  .github/workflows/codeql.yml
 a47735f905b1463c148d0b71f5db852beec873782be18b6d1e7a55ec54a1baf4  .github/workflows/doc-quality.yml
-592a86e79d00a0bbbaed2e20d2fba91aba6744baa4cfb8222ea41fe422543fce  .github/workflows/e2e-integration.yml
+1aa99d1c75d090fab98411c98319bad4b670f70cf6be6c701a0fd82216103655  .github/workflows/e2e-integration.yml
 43194d2f4ce1c37f70b4db099262149a4cd28f541eaa363f73adf05801b89935  .github/workflows/harness-hash-verify.yml
 b0302fc487734c9d925accc83f58ac86587eb991ec35d675c43282aebf295ecf  .github/workflows/leading-indicator-watch.yml
 87e8277670edd32734871ea373f37395329a19a572e9ccc39a9af3789087ecff  .github/workflows/lint.yml

CHANGELOG.md

@@ -7,12 +7,6 @@ versioning follows [SemVer 2.0.0](https://semver.org/spec/v2.0.0.html).
 
 ## [Unreleased]
 
-### Changed
-
-- Refresh the neutral end-to-end convergence harness to validate against the
-  published `@intentsolutions/core@0.10.0` kernel while preserving the same
-  five-repository seam assertions.
-
 Post-0.3.0 work: continuous-spec-compliance / Spec Authority Kernel (SAK) skeleton, the 6-contract upstream deep-capture program, the spec-drift upstream-surface registry, governance hash-chaining, the IEP PR-review-gate policy, CI-integrity false-green fixes, the tag-triggered release evidence lane, and the end-to-end 5-repo integration test framework.
 
 ### Added
@@ -128,6 +122,9 @@ skill evals still hit the judge-truncation (#175) + reasoning-model
 
 ### Changed
 
+- Refresh the neutral end-to-end convergence harness to validate against the
+  published `@intentsolutions/core@0.10.0` kernel while preserving the same
+  five-repository seam assertions.
 - Blueprint B § 7 — add Parallel Change discipline (expand-contract for kernel schema additions) (#148)
 - Scope advisory doc-lint to exclude machine-generated research results (#157)