No lint validates deferred:integration (STORY-NNN) pointer citations against the owning story

## Process-gap

Code and test artifacts carry deferral-pointer comments of the form `deferred:integration (STORY-NNN)` that name the future story which will complete a deferred seam. When a re-scoping decision moves a deferred seam from one story to another, the pointer must be updated at every citation site. There is no lint or check that validates these pointers, so stale pointers survive.

Observed pattern: a re-scope updated the owning story for a deferral seam, but the fix-list enumerated only a subset of the sibling files carrying the citation. Stale pointers then surfaced one-at-a-time across multiple subsequent review passes (3 distinct instances of the same propagation class in one wave), each costing a fix + re-review cycle. A future owner grepping for the (now-wrong) story ID gets misdirected.

## Proposed mitigation

Ship a lightweight check (CI lint or convergence-gate step) that: (1) extracts all `deferred:integration (STORY-NNN)` citations, (2) for a given re-scope, confirms no residual citations name the old story on the re-scoped seam, and (3) optionally cross-checks each cited STORY-NNN exists in the story index. Alternatively, require the re-scope determination memo to enumerate the exhaustive citation-site list (grep-derived) rather than a hand-picked subset.

## Severity
Low-Medium — comment/documentation drift, no runtime impact, but recurring and cycle-costly.