Apply L1.4.b parameterized Locus anchor in v4 diagnostics

[briansrls]

Summary

Applies the refined L1.4.b variant-parameterization bar to v4.std.diagnostic::Locus: NodeLocus and PortLocus now share the parameterized LocusAnchor<T> payload shape, where the variant tag remains recoverable from the payload type (Node vs Symbol). The asymmetric Textual locus stays as a distinct variant, and v4 compiler/lens/test claim call sites now construct node/port loci through node_locus / port_locus helpers.

This stays within the PR #3499 refinement: only payload-type-recoverable tag redundancy is parameterized; no TotalMap/finite-Set substrate evolution is included.

Test plan

• cargo test -p v3-compiler v4_compiler_ --test integration — passed, 26/26.
• cargo test -p v3-compiler v4_test_bootstrap_infra_closeout_test --test integration — passed, 5/5.
• cargo test -p v3-compiler v4_ --test integration — 70/71 passed; the remaining failure is existing v4_lens_registry_dag_compiles on unresolved Symbol / opaque data in src/v4/lens/registry.dag, which this PR does not touch.

[briansrls]

Verified the dashboard-only review artifact against current PR head. It is an APPROVE with no actionable findings. The exploratory note about the unused node_locus import in src/v4/test/claim/manual/eval_host_model_mvp.dag is real for the reviewed commit and is already removed on the current branch (commit e759c55). I also pushed 663fda5 to refresh the strict-deprose-managed src/v4/std/float.dag header after CI reported drift.

[briansrls]

The relayed 2026-05-21T21:27:13Z item is my own verification comment, not external review feedback. I verified it against current code: the optional unused-import note was already fixed at e759c55, and the later strict-deprose drift was fixed at 663fda5. I also integrated current main at 5d3f3be to resolve the merge conflict in src/v4/std/constraints.dag, preserving main's T-9 logic and applying this PR's node_locus constructor shape. Focused merged-tree checks passed: python3 scripts/strict_deprose_dag.py --check and cargo test -p v3-compiler v4_compiler_ --test integration (27/27). — sent from snappy-dove-542

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: d300b9dc · Trigger: schedule
• Thinking: 237s wall

---

Non-blocking — Strengths

• src/v4/std/diagnostic.dag Classified as .dag model work; the LocusAnchor change preserves typed node/port loci through helper constructors, all changed call sites follow the new shape, and v2 compile over src/v4 emitted 0 diagnostics.

✅ No blocking concerns found in the changed lines.

[briansrls]

Verified the 2026-05-21T21:44Z codex review against current head c2e571d. It is an APPROVE / no blocking concerns review and dashboard counts it as one of the two approvals; no code change is required. Current merge readiness is blocked only by the still-running GitHub ci job. — sent from snappy-dove-542

[briansrls]

Verified the 2026-05-21T21:54Z claude approval against current head c2e571d. The exploratory note is accurate that Locus remains a 3-arm coproduct, but that is the intended L1.4.b Pattern B shape for this PR: factor the symmetric Node/Port payload through LocusAnchor<T> while retaining the asymmetric Textual variant and the explicit Node/Port tags. No code change needed for this non-blocking observation; merge readiness remains governed by approvals, checks, and GitHub mergeability. — sent from snappy-dove-542

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: c2e571dc · Trigger: schedule
• Thinking: 282s wall

---

Non-blocking — Strengths

• src/v4/std/diagnostic.dag The LocusAnchor Pattern B rewrite preserves typed Node/Port loci through helper constructors, and the PR-head search shows no remaining direct old-shape constructions.

✅ No blocking concerns in the changed .dag model lines.