docs(scripts): the dangling-citation detector's header claimed no workflow runs it

[wshallwshall]

PR 560 (b47c9fd) added tests/test_dangling_citation_check.py to ci.yml's DOC_GUARDS, and left
the module docstring asserting the opposite. Measured at origin/main: the path now occurs once
under .github/, where the header said ZERO, with controls in the same pass (backlog_status_check
3, ledger_check 2, scan_forbidden 3, an invented path 0).

Three things the old text got wrong, not one:

• the count is stale -- it is 1, not 0
• the wiring claim is incomplete -- the test now reaches CI by the docs lane as well as the tooling
  leg, and the sole .github hit is the TEST, not the script
• the grep warning has inverted. It told a reader to grep the full name because a bare
  citation_check hits a workflow belonging to the sibling. A test filename contains its subject's
  name, so the full name now hits ci.yml too. The discriminator is script versus test, not full
  name versus bare.

Also corrects a stale in-header measurement (26 of 32 -> 41 of 50, floor 1352, re-run against the
tree) and records that the tooling leg IS pinned -- by a glob over tests/test_*.py in
test_tooling_partition.py, not by name, so grepping for this module finds no guard and the leg
reads as unpinned when it is not.

Docstring only: the module AST with the docstring removed is byte-identical before and after,
verified with a control that detects a mutated function name. ruff/mypy strict clean. The full
suite's 29 failures are identical with and without this change, compared as sets rather than counts.

The item citation is omitted from the subject deliberately: BACKLOG #1235's claim is held by a
worktree that no longer exists, and whether this change closes that row is left to Dispatcher's
disposition rather than presumed here.

Built by Cleaner, verified and opened by Lander.

🤖 Generated with Claude Code

[wshallwshall]

Update since this PR's commit body was written: #1235 has confirmed CLOSED (parse_items at origin/main, verified by Cleaner and independently consistent with my own PR 587 closure earlier tonight). The commit's "whether this change closes that row is the Dispatcher's disposition call" line describes a question that's since been settled -- leaving the commit as-is (not amending a pushed sha for a stale deferral) but noting the resolution here for anyone reading this PR.