Retarget the UK build to FRS 2024-25: re-pin raw vintages, regenerate parity instruments, re-measure gate baselines #723

Workstream E of the UK-migration master epic #665, child of #145. Starts as soon as #681 (E5) and #682 (E6) land; prerequisite for verifying #684 (E8) at the right vintage and for #686 (E10) parity.

Decision recorded (2026-08-19)

The migration targets FRS 2024-25 directly (survey/base year 2024, calibration year 2025) — the incumbent's live production line (enhanced_frs_2024_25, v1.56.x). The 2023 vintage build is deferred (#687 rescoped): parity at a vintage consumers have already moved past would buy a second parity pass later for no independent value. Donor-survey vintages are unchanged — the incumbent's 2024-25 build uses the same SPI 2022-23, WAS round 8, LCFS 2023-24, and ETB files already staged locally; only the FRS spine and the target year move.

This child is deliberately mechanical — the tooling for every step already exists; the work is re-pointing and re-running, plus adjudicating what the re-runs surface.

Scope

1. Raw-vintage audit at 2024-25 (the E2/U7 pattern): sha+size-pin the FRS 2024-25 tab set (25 tables, staged locally), and confirm the stage declarations accommodate the known 2024-25 schema quirks the incumbent handles explicitly — the household table is not sernum-ordered (the weight-scrambling sort bug), the salary-sacrifice asked-state fields, and any EDUCQUAL/derived-field deltas surfaced by the audit.
2. Re-pin stage manifests + adopt the release-object discipline (moved forward from the old E11): one frozen declarative release object naming survey/base/calibration years, UKDS study number, DOI, zip sha; tape years labeled by tax year (2024_25) per the SPI-stage convention.
3. Regenerate the parity instruments from enhanced_frs_2024_25 (pull + pin revision and sha; requires HUGGING_FACE_TOKEN): efrs_parity_reference, the known-gaps register (re-adjudicated — 2023 gaps may close, new ones may appear and must be signed, not inherited), and the record-count identity re-derived at the new vintage.
4. Re-verify the landed stages (E1–E7) at the new vintage: re-run each per-stage verification harness against the re-frozen bar; land fixes for any vintage-sensitive declaration.
5. Re-measure the weighted-integrity baselines at the new vintage (tools/measure_uk_weighted_integrity_baselines.py): weight-ratio reviewed maximum, ESS fraction, zero-weight strata, and the input-mass reference. These become the candidate thresholds Assemble the microcosm-built UK spine: whole-spine parity and swap acceptance #686 arms.

Out of scope

• Target period alignment to calibration year 2025 — lives with the contract (Adopt the UK national calibration contract (re-homed from chronicle#164 per the #166 ruling) #707) and target references (Consume Ledger UK facts: uk_target_references, mapping module, and a compile-parity gate #622).
• The calibration-year materialization mechanism (dataset uprating 2024→2025 vs period-aligned facts) — adjudicated with Adjudicate the UK spine grain at the ingest/calibration seam #688; whichever way it lands, it is a thin WS-F slice, not this child.
• The swap itself (Assemble the microcosm-built UK spine: whole-spine parity and swap acceptance #686).

Acceptance

• The full ingest StagePlan runs at smoke rung on the 2024-25 tabs.
• Regenerated parity instruments committed, pinned to the incumbent revision/sha they were extracted from.
• Per-stage verification green for E1–E7 at the new vintage; every divergence either fixed or signed in the known-gaps register.
• New-vintage integrity baselines recorded and referenced from the Assemble the microcosm-built UK spine: whole-spine parity and swap acceptance #686 scope.