F1a: repoint deleted-TASKS.md cites in test/claim + v3 test comments to live concept issues

[briansrls]

What

Comment-only PR (3 of 3 in the F1a lane; siblings #4767, #4768). Repoints every TASKS.md cite under src/v4/test/ and in src/v3 test/module comments to its dissolve-on concept tracking issue (#4757–#4766 family; mechanism per docs/planning/label-hygiene-census-2026-06-12.md, PR #4749):

• v4 claim eval runner: Outcome-carrying eval replaces structural projection and Outcome→Bool claim bridges (re-anchor of deleted TASKS.md T-22 binds) #4757 claim eval runner — the T-22-outcome-to-bool-claim-bridge marks (claim_pipeline ×4, parse wave2a ×3, grammar claims ×6, sg_claims, manual_corpus_roster) and the W-T-10/W-T-11 Outcome-eval marks in the five mvp1_*_add_translate.dag files
• coproduct-arm reflection replaces hand-enumerated variant projections (re-anchor of deleted TASKS.md T-3/T-19 binds) #4759 coproduct reflection — generated/coproduct_exhaustiveness.dag (3), generated/algebra_law_conformance.dag
• infer-authored grounding replaces hand-enumerated fixture facts (re-anchor of deleted TASKS.md T-9/T-10 binds) #4760 infer grounding — infer_algebra_ref_grounding_anchor.dag, fixture-grounding marks in the translate files
• lens application + testgen canonicalization: CompileLens locals, coverage-key projections, LeafModelFixture rows, toolchain coproducts (re-anchor of deleted TASKS.md T-4/T-19/T-22-follow-up binds) #4762 lens/testgen — generated/testgen_category_wishlist.dag, coproduct_exhaustiveness.dag:132 (M2 fold reflection)
• typed DagRoundTripWave1Readiness receipt replaces conjunctive Bool round-trip gates (re-anchor of deleted TASKS.md T-36 binds) #4763 round-trip readiness — dag_ingest_round_trip.dag anchor (former §T-36)
• TestClaim corpus eval workflow gates: zero-Fail/zero-Deferred CorpusEvalReport (re-anchor of deleted TASKS.md T-38 binds) #4765 corpus eval — workflow/manual_corpus_eval.dag, workflow/testclaim_corpus_runner.dag (former T-38)
• v3 comments: sg0_census_test.rs P5 receipts (T-22/T-38 → v4 claim eval runner: Outcome-carrying eval replaces structural projection and Outcome→Bool claim bridges (re-anchor of deleted TASKS.md T-22 binds) #4757/TestClaim corpus eval workflow gates: zero-Fail/zero-Deferred CorpusEvalReport (re-anchor of deleted TASKS.md T-38 binds) #4765), v4_emit_host_harness_test.rs, v4_hollow_alias_gate.rs (T-30 ledger paragraph → 'former'), v4_t15_self_host_fixed_point_harness_test.rs (§T-15 prose), boundary/english_ingest_fail_closed.dag, dead-URL anchor dropped in fact_density_anchor.dag

Coupled test literal

v4_test_bootstrap_infra_closeout_test.rs:184 asserts the verbatim wishlist mark text — updated here in the same PR. Line 584 (the verification.dag twin) is deliberately NOT touched here: that file's mark is rewritten by #4767, which carries its literal update. The two PRs are independent and squash-merge in either order.

Coverage note

No skip-listed files are touched (none of #4741/#4747/effect-emit-PR1's files have TASKS.md cites under test/). tools/ci_affected_components keeps its src/v4/TASKS.md path-matcher test string — it tests that .md paths don't affect v4, and any path string works there; left as-is.

Verification

grep -rn 'TASKS\.md' src/v4/test/ src/v3/ --include='*.dag' --include='*.rs' now returns only re-anchor/former/supersedes annotations plus the intentionally-deferred line 584. Diff is comments plus the one coupled assert literal.

🤖 Generated with Claude Code