backlog: file #1315 -- prose path:line citations carry no token, so nothing can verify them

[wshallwshall]

Replaces #509, whose head had no path forward. Same item, re-applied cleanly onto current main by its author.

Why the branch changed rather than the PR merging

#509 needed main merged in after #1314 landed at the BACKLOG tail. That merge turned out to be closed from every tree, by two gates that each behave correctly and together leave no route:

• the worktree gate refused to switch the allocating tree onto the branch — it saw that tree busy on another branch and read the swap as a hijack
• a fresh worktree would then have failed the ledger gate, which keys ownership on the allocating path

Closed in opposite directions. COMMON names the way through and the author took it: git switch -c <new> origin/main from the allocating worktree — same worktree, so the ledger claim still holds; new branch, so nothing looks like a hijack.

That deadlock is worth knowing independently of this item. Neither gate is wrong; the interaction is what has no exit, and the escape is non-obvious from inside either one.

The item

• 2,090 scorecard anchors carry an expect token the tree confirms.
• 3,543 occurrences / 2,871 distinct prose citations carry a bare path:line and nothing else.

Not rot: measured decay in the 1107-1199 half is 16 of 1,196 distinct pairs. Nothing can tell whether a citation has gone stale, because a bare line number makes no claim anything could check. A detector built for rot returns zero and the zero means nothing.

The clause at its centre grades rather than fails: not "is there a line number" but "can something else in the sentence find the line again" — which turned raw counts of 68 / 8 / 3 across five seats into naked counts of 24 / 1 / 1.

Verification

Author's numbers, independently re-checked here against origin/main at 664443f8:

• 1 file, 15 insertions, 0 deletions; 2 commits; zero behind main
• 328 items on main, 329 here; #1315 added, nothing removed
• zero banner-set changes on any existing item; zero items carrying both an open and a closed banner
• SHIPPED-BUT-OPEN count unchanged at 8, so main's own notes survived
• author resolved on \r\n boundaries and asserted bare LF 0 before and after, rather than asserting a line count — the line count legitimately drops by the marker lines
• the prose-quoted conflict marker at ~11258 was used as the positive control and confirmed still present

The author verified the ledger gate fires on this branch rather than assuming a silent pass: cherry-pick --continue printed no hook output, so they ran ledger_check.py directly (exit 0) and then made an empty probe commit to confirm the hook actually runs, dropping it with reset --soft.