fix(pm): drop the no-path-population marker the widening-tell gate cannot honour

The gate's CI command is its own --self-test, so on the marker's own second
listed cause the declaration read true. dispatch-gates' live guard refuses it
anyway the moment a family NAMES paths, and this one names 59: 9 from its module
body (the registry table, the objectui mirror glob, two repo slugs, four fixture
filenames) and 50 inherited from the two registers it imports on purpose,
SUSPECT_TIER_GLOBS and REGEN_ARTIFACTS.

Getting those out of a scanned position would mean hand-copying two registers
this file imports precisely so it can never disagree with them - the drift
check:pm-governed-prose exists to stop, and this gate's strongest property. So
the marker goes and the derivation stands. The header now records what the
derivation says, which part of it is right (packages/spec/src/**,
api-surface/**, the three registries - the surfaces this gate polices) and which
part is noise inherited from REGEN_ARTIFACTS, and why one cheap self-test in a
MATCHED column is the smaller error than a marker sitting above a live
population.

The case that catches this sits at ~1534 of dispatch-gates' self-test and needs
well over 540s to reach, which is why three local runs missed it.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018dxq7YqsLDMeZDZ5AzsgJX

Author: claude

scripts/pm/check-widening-tells.mjs

@@ -158,7 +158,47 @@
  * diff, and failing an unrelated PR's CI over it would punish the wrong actor.
  */
 
-// dispatch-gates: no-path-population -- this gate reads no file in the tree as its input: it judges a PR's DIFF, supplied by its caller from the API or from `git diff`, so no card's file surface can predict it. The path globs in its module body are patterns matched against diff filenames, never files it opens (#16448)
+// ⛔ NO `dispatch-gates: no-path-population` MARKER HERE — deliberately, and the
+// reasoning is worth the paragraph because the marker LOOKS right.
+//
+// This gate's CI command is its own `--self-test`, which is the second of the
+// three causes that marker's docblock lists ("the derivation NEED NOT place it").
+// On that reading the declaration is true: no card's file surface should
+// schedule this command, because running it says nothing about that card's diff.
+//
+// But the marker is refused by `dispatch-gates`'s live guard the moment a family
+// NAMES paths, and this one names 59 of them (measured on this tree): 9 from its
+// own module body — the registry table, the objectui mirror glob, the two repo
+// slugs and four fixture filenames — and 50 inherited from the two registers it
+// imports ON PURPOSE, `SUSPECT_TIER_GLOBS` and `REGEN_ARTIFACTS`. That guard's
+// own text gives the fork: "If the literals are the real population, delete the
+// marker and let the matched column do its job; if they are artifacts rather
+// than a population, the marker stands and the literals do not belong in a
+// scanned position."
+//
+// Both halves of getting them out of scanned positions cost more than they buy:
+// the 9 would have to become segment predicates instead of paths, and the 50
+// would have to become a HAND COPY of two registers this file imports precisely
+// so it can never disagree with them — which is the drift
+// `check:pm-governed-prose` exists to stop one family over, and the single
+// strongest property this gate has. So the marker goes and the derivation
+// stands.
+//
+// What the derivation now says, and why it is not wrong: a card touching any of
+// those 59 paths gets `pnpm check:pm-widening-tells` in its MATCHED column. For
+// `packages/spec/src/**`, `packages/spec/api-surface/**` and the three
+// registries that is exactly right — they are the surfaces this gate polices,
+// and a dev editing one is the dev whose claim it will judge. For the tail
+// inherited from `REGEN_ARTIFACTS` (`*/test-typecheck-debt.json`,
+// `content/docs/references/**`, and the rest) it is noise, and the cost of that
+// noise is bounded and small: the command is an offline self-test that runs in
+// about a second and whose green means "the tells still work". An
+// over-matched gate pastes one cheap command into a prompt; the alternative was
+// a marker sitting above a live population, which is the rot direction
+// `dispatch-gates` measured and refused. ⛔ Do not re-add the marker without
+// first removing the imports — a green local run is not evidence, because the
+// case that catches this sits at ~1534 of the self-test's assertions and needs
+// well over 540s to reach (#16448 patch round 2).
 
 import process from 'node:process';
 import { existsSync, readFileSync } from 'node:fs';