test(readme): the selftest numbers rot loudly, against the suites' own summaries

Order 20260905T053602Z-6743615726-from-.claude (.claude, 2026-09-05), asking
for the check this repo recommended when it re-measured the README a run
earlier. The badge and the five `## Development` comments rotted twice in a
row - 529 carried from 0.25.0, then a badge saying 868 beside comments summing
to 792, two different wrong sums of the same fact on the same screen - because
nothing compared them to anything.

It lives in tests/selftest.test.mjs, not in a bash suite, and the choice was
measured rather than assumed. The order's parenthetical pointed at whichever
suite already pins README/catalog invariants; no such suite exists -
`grep -ln README scripts/*selftest*.sh` returns board-selftest.sh alone, on two
incidental hits (a prose comment and a research/README.md fixture). This
wrapper is the only place where all five numbers exist at once in a run that
already happens. runSuite() captures each suite's own summary line, so the
truth source is the line the suite prints. A check inside one suite could see
its own total but would have to RE-RUN the other four - 212s sequentially,
measured 2026-09-05 under /bin/bash 3.2 (coord 16, board 169, route 12,
orders 5, guard 10) - and grepping `check` calls out of the scripts is both the
second copy of the counting the order warned against and a wrong one, since
those calls sit inside loops.

Three properties are deliberate. The badge is compared against the MEASURED
sum, never against the five README comments: a badge agreeing with five stale
comments is the 868-beside-792 shape one layer down. A suite that stops
printing a recognisable summary FAILS the check rather than being skipped -
an absent measurement must not read as a matching one. And the check adds no
bash check anywhere, so the five counts and the 893 badge are unchanged by its
arrival, exactly as the order expects; a counted self-check would have had to
compare against PASS+FAIL+1 and would break for whoever adds a check after it.

Ground truth on this HEAD, run before anything was written: coord 257,
board 393, route 73, orders 116, guard 54 = 893, 0 failed in all five - the
README was already correct, so the red step is the mutation. Mutation-verified
in both directions: 73 -> 74 on the route comment gives "README says
route-selftest has 74 checks; it reported 73"; 893 -> 894 on the badge gives
"README's badge says 894 selftest checks; the five suites reported 893";
restored, npm test is 12/12 green.

Bounded gap, stated rather than closed: CLAUDE.md's own copies of the five
counts are a second surface carrying the same numbers and are NOT checked.
Measured and left alone - widening the check to it was not ordered.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

README.md

@@ -200,7 +200,7 @@ Note that raising the inbox's priority (Rule 7) deliberately does **not** widen
     bash scripts/route-selftest.sh              # 73 checks, incl. the route->board round trip
     bash scripts/orders-selftest.sh             # 116 checks, incl. the 20-way barriered claim race
     bash scripts/state-line-guard-selftest.sh   # 54 checks, incl. the Edit replace_all projection and the ratchet
-    npm test                                    # all five selftests plus the hook tests, via node --test
+    npm test                                    # all five selftests, the hook tests, and the README-number check
 
 TDD is the house rule: every behavior change lands with a failing selftest check first.