Replace stripPromotedUrls phantom-entry scrub with a section-ownership/provenance model

[s-annam]

Follow-up from the #127 segmentation spike (docs/segmentation-spike.md), under epic #109. PR 3 — depends on SectionedResume (Follow-up A).

Goal

Replace the stripPromotedUrls phantom-entry scrub (openresume.ts:133, driver :196-:250) with a provenance/ownership model: a line consumed by the contact extractor (a promoted identity link) is marked consumed and never offered to extractProjects/extractAchievements/etc.

Why

Because extractContact matches LinkedIn/GitHub document-wide (anywhereOnDoc, extract-fields.ts:455), a footer "Links" line is promoted into the contact card AND survives as a phantom project/achievement entry. Today that duplicate is scrubbed after the fact by slug subtraction across every section body. Better segmentation alone does NOT delete this (spike §1.3) — the same URL is legitimately both contact and a links-section line. The right fix is ownership: the contact extractor claims the links it consumes so they never re-render.

Scope

• Add a "consumed lines" / ownership signal to the section consumption flow so a link claimed by contact extraction is removed from candidate pools for other extractors.
• Retire stripPromotedUrls, isPromotedUrl, PROMOTED_LABEL_RE, urlSlug-for-stripping, and the per-section strip driver in buildHeuristicResult.
• Use the fix(heuristics): Word-template parsing + identity-link, education & experience extraction #125 identity-link fixtures as regression anchors.

Acceptance

• No identity link renders twice (contact + body) on the fix(heuristics): Word-template parsing + identity-link, education & experience extraction #125 fixtures.
• Corpus goldens audited; any movement explained (medium-risk PR, touches contact + section assignment).

Refs #127, #109.

[s-annam]

Implementation Complete — 77e6d03 (PR #140, pending merge)

Implemented in stacked PR #140 (base: refactor-137-132-fallow-sectioned / #138, which provides the typed SectionedResume this builds on).

Acceptance Criteria

• Consumed-line ownership signal — extractContact returns consumedLines: ReadonlySet<PdfLine>; buildHeuristicResult runs name+contact first, then stripConsumedLines removes owned lines from body pools before extractProjects/extractAchievements/etc. A claimed identity link is removed from candidate pools, so the phantom block is never created.
• No identity link renders twice on fix(heuristics): Word-template parsing + identity-link, education & experience extraction #125 fixtures — the 3 promoted-link de-duplication tests pass; bare URL lines consumed, deeper-path / longer-handle bullets survive (not owned).
• Retire stripPromotedUrls / isPromotedUrl / PROMOTED_LABEL_RE / urlSlug-for-stripping / per-section strip driver — done; only doc-comment refs remain.
• Corpus goldens audited — zero movement; all 24 snapshots pass unchanged.

Notes

• Ownership keyed on PdfLine referential identity. Owns only pure identity-link lines (label + URL only); mixed lines keep their non-URL remainder — faithful to the targeted footer/Links-line bug, confirmed by zero golden movement.
• Stale refs (openresume.ts:133, extract-fields.ts:455) remapped to the post-refactor: decompose scan/extractSkills + typed SectionedResume (#137, #132) #138 locations (openresume.ts buildHeuristicResult, extract/contact.ts).
• Validation: typecheck clean, 616/616 tests.