fix(scripts): render an unread clause-② carrier as NOT MEASURED, not as absent

`declarationFromPullRequest` rendered every non-matching label list the same
way — ``carrier: `needs:contract-review` is not on this PR (N label(s) read)``
— including N = 0. A zero-label read is not an absence: the event payload is a
snapshot taken when the run fired, so an empty list is equally consistent with
"this PR carries no labels" and "nothing had been labelled yet", and nothing
inside the payload separates them. This file already owns the right word for
that state and uses it verbatim in its own payload block ("the `opened` run at
21:40Z would have read NOT MEASURED"); it just did not spell it here.

The string was not the damage. `renderLevel`'s `enforce` message printed the
unread carrier beside the `Clause-②:` line under "The two declarations
disagree", and then offered remedy 2 — "correct it at the producer: the
carrier is the review seat's to place and to clear". Pointed at a carrier the
run never read, that is an instruction to strip a review requirement in order
to clear a red whose real cause was remedy 1. This file's own text says the
two remedies are not interchangeable.

So the reader now returns a three-valued `carrier` axis (`present` / `absent`
/ `not-measured`) beside `value` and `arm`, `judgeLevel` carries it through,
and `renderLevel` softens the header and remedy 2 only on a positive
`not-measured`. A measured absence keeps the original verdict, the original
routing and the original bytes; every verdict keeps its exit code.

⛔ The payload boundary is untouched: no API call, no token, no live label
read. The `labeled`/`unlabeled` triggers remain the mitigation.

A new self-test battery pins both directions from one harness, plus the
fixture control the card asks for (a PR carrying the carrier must report it
PRESENT).

Claude-Session: https://claude.ai/code/session_017ef78bLdybu3AffehKkhfk
Co-authored-by: Claude <noreply@anthropic.com>

Author: claude