workflow-fix: extend pre_reg head-noun list for the #1586 'registered layer' escape

Overview / Motivation

Auto-filed by the workflow-fix-on-bug protocol from a prose follow-up raised by the clean-result-critic on task #1586 (emitting agent: clean-result-critic, round 1, 2026-07-23).

Goal

Extend audit_clean_results_body_discipline.py's pre_reg head-noun list to catch the #1586 escape class ("the registered layer" ×5), measuring the critic-sketched sibling nouns under the file's established corpus-measurement discipline.

Workflow gap

• Bug observed: five "the registered layer" Lens-7 register violations in #1586's clean-result body passed the audit's pre_reg mechanical check; only the LM critic caught them.
• Why it is a workflow gap: the pre_reg branch (added #1419, extended #1475) covers a measured head-noun list that lacks layer — the exact escape-then-extend pattern both prior fixes followed.
• Confidence (emitter): high (probe-verified)
• verified-at-filing: uv run python -c "re.search(PATTERNS['pre_reg'][0], 'the registered layer')" → False ('the registered verdict' → True; live probe against the current module, 2026-07-23). grep -c '\blayer' scripts/audit_clean_results_body_discipline.py → 0. Context read per clause (c): the existing #1419/#1475 check is present but its head-noun list demonstrably excludes this class — a DISTINCT gap, not a landed fix.

Proposed change (candidate diff sketch — refine in planning)

• Add layers? to the pre_reg head-noun alternation; MEASURE (per the file's corpus discipline: full-pattern old-vs-new match-start diff over all promoted/parked bodies) the critic's sketched siblings lattice|rungs?|windows?|bands?|reads?|margins? and include only those with 0 (or explicitly accepted) benign verb-use false positives — margin/read are the likely FP-risk cases to measure carefully.
• Update the pattern's provenance comment with the #1586 incident strings ("the registered layer", "reported at the registered layer").

Scope / surfaces

• Primary target: scripts/audit_clean_results_body_discipline.py
• Grep the workflow surface for the pattern before editing and update every hit; list them in the plan.

Constraints / invariants

• Workflow-surface only. scripts/workflow_lint.py --check-asks passes; ruff passes; the file's measured-corpus discipline (0 unaccounted FPs) is the acceptance bar, matching #1419/#1475.
• This session runs under EPM_WORKFLOW_FIX_SESSION=1 and carries a workflow_fix_target: Provenance line — it MUST NOT auto-route its own subagents' workflow-fix candidates.

Provenance

• workflow_fix_target: scripts/audit_clean_results_body_discipline.py
• fingerprint: 4a643434c286

Surfaced prose (verbatim from the clean-result-critic round-1 report): "mechanizable: yes — audit regex \bregistered\s+(layer|verdict|margin|read|lattice|rung|window|band)\b over the four content sections (…the sketch is in the marker for routing)."