fix(workbench): reconcile traceability for implementing-domain-driven-design and refactoring-guru

[rxavier1979]

Refs #1. Blocks #6 — rule IDs cannot be surfaced from a mapping that does not reconcile.

Status: diagnosed 2026-08-20, not started.

Verified: 12 of 14 books reconcile exactly. Two do not, and the mismatch is bidirectional in both.

Not verified: which specific bullets lack IDs and which IDs are orphaned. The sweep counted; it did not diff line-by-line. The first task on this issue is that diff.

The central fact

Book	Section	Traceability IDs	Released bullets	Delta
implementing-domain-driven-design	Decision rules	18	19	1 bullet has no ID
implementing-domain-driven-design	Trigger rules	12	10	2 IDs are orphaned
refactoring-guru	Decision rules	24	24	OK
refactoring-guru	Trigger rules	20	14	6 IDs are orphaned

Orphaned IDs mean rules were removed from mini.md without updating traceability.md — exactly the drift PROCESS.md "Traceability Rules" exists to prevent.

These are the only content-level defects found anywhere in the corpus. Everything else that failed a check is mechanical (line endings, symlinks) or absent (NOTICE, self-governance files).

Goal

Every released mini rule maps to exactly one M* id, and every M* id maps to exactly one released rule, in all 14 books.

Decisions (settled 2026-08-20)

#	Decision	Rationale
1	Diff before editing — enumerate the unmapped bullets and orphaned IDs explicitly, and post them as a comment on this issue before any file changes	The counts say that it is broken, not what is broken. Editing on a count is guessing.
2	An orphaned ID is retired, never renumbered	PROCESS.md treats IDs as stable references. Renumbering silently invalidates every external citation. Record as retired (rule removed <date>).
3	An unmapped bullet gets the next unused id, not a gap-filling one	Same reason. implementing-domain-driven-design currently tops out at M28, so the new one is M29.
4	This is a book-specific miss, not a process bug	PROCESS.md already required reconciliation; the process is correct and was not followed. Per its own "Process vs Book Diagnosis" section, that means fix the books and do not amend PROCESS.md.

Acceptance

• the specific unmapped bullets and orphaned IDs are enumerated in a comment before any edit
• 14/14 books reconcile: decision-IDs == decision-bullets, trigger-IDs == trigger-bullets
• retired IDs are recorded as retired, not reused
• rules_lint.py checks 1 and 2 catch a reintroduced mismatch
• decision-log 0004 — Rule IDs are stable references: orphans retire, they never renumber

Reproduction

# per book: count unique `M\d+` in _rule-workbench/<book>/traceability.md,
# split by the "Decision rules:" / "Trigger rules:" subsections,
# and compare against "- " bullets under the matching H2 in <book>/<book>.mini.md

The sweep script that produced the table above is attached to #5.