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

changelog.d/723-uk-frs-2024-25-retarget.changed.md

@@ -0,0 +1 @@
+Retarget the UK build to FRS 2024-25 (#723): a new frozen release object (`uk/frs_release.json` — survey/base 2024, calibration year 2025, SN 9563, DOI, UKDS zip sha, HF acquisition pins) drives lockstep asserts over the re-pinned raw-tab manifest (all 21 `frs_table` artifacts at sha+size for the 2024-25 tabs, SPI-convention `tax_year_start`/`ukds_study_number`/`doi` keys), the build period moves to `"2024"`, the take-up contract's `build_year` follows (tax-free-childcare and TV-evasion rates flip to their 2024 date keys), and the HMRC 2023-24 published surface is re-mapped as a signed nearest-available-vintage declaration (`period_mapping: "latest_published_tax_year"`, values and vintage labels unchanged — no uprating). The parity bar re-freezes on `enhanced_frs_2024_25.h5` @ the v1.56.14 tag: the eFRS parity reference (145 columns, entity counts (16,288 raw + 10,000 SPI) × 2 + 270 CGT band donors = 52,846 — now test-pinned), the release-input coverage manifest, and the input-mass reference registry identity all move to the 2024-25 pins; gate thresholds do not move (#686 arms the re-measured baselines). Six stages' SN 9252 (FRS 2022-23) source citations were a standing prose defect, fixed and signed.

changelog.d/723-uk-ingest-scale-ladder.added.md

@@ -0,0 +1 @@
+The UK ingest spine driver joins the #627 scale ladder (#723): `--sample-fraction` (0.01/0.10/1.0) subsamples households right after the raw `frs_spine` stage through the generic frame-sampling helpers, records a sampling receipt and rung token in the build sidecar, refuses `--checkpoint-dir` on sampled rungs, and adopts the named-edge abort receipt. Sampled rungs get an explicit receipt posture on three full-scale fences — raw-tab person coverage restricts to survivors, the SPI synthetic stack scales by the fraction, and the ultra-sparse effective-mass restoration check records instead of raising — while f100 keeps every fence strict.

changelog.d/723-uk-was-bridge-donor-pin.fixed.md

@@ -0,0 +1 @@
+The WAS round-8 bridge-donor artifact in `lcfs_consumption` declared a different `locator` than `was_wealth`'s pin for the same tab, so the full-roster spine driver refused every licensed run at the pin-coherence check — a break invisible to secrets-free PR CI (#723). The locator is aligned to the caller-supplied convention and a hermetic regression test now collects `_artifact_pins()` over the complete manifest.

packages/microcosm-build/src/microcosm/build/spec_engine/schema/sources.schema.json

@@ -5309,6 +5309,15 @@
       },
       "runtime_sha256_required": {
        "type": "boolean"
+      },
+      "tax_year_start": {
+       "type": "integer"
+      },
+      "ukds_study_number": {
+       "type": "integer"
+      },
+      "doi": {
+       "type": "string"
       }
      }
     },

packages/microcosm-build/src/microcosm/build/uk/cgt_source_stages.json

@@ -35,8 +35,8 @@
             "3_3_2021-22",
             "3_4_2020-21"
           ],
-          "mapped_build_period": 2023,
-          "period_mapping": "tax_year_start",
+          "mapped_build_period": 2024,
+          "period_mapping": "latest_published_tax_year",
           "basis": "Individuals with a CGT liability only; gains after losses and attributed gains but before the annual exempt amount; taxable income after reliefs and the Personal Allowance; trusts excluded; counts publish in thousands and amounts in GBP millions.",
           "runtime_sha256_required": true
         },
@@ -93,7 +93,7 @@
           "suppressed_cell_allocation": "count implied by the cell's published gains at the band-total mean",
           "column_reconciliation": "every income column rescales onto its published All-row taxpayer total",
           "shortfall_policy": "proportional scale-down when the population holds less gainer mass than published taxpayers",
-          "minimum_allocation_people": 1.0,
+          "minimum_allocation_people": 1,
           "weights": "household_weight mapped to persons; no person splits across bands"
         },
         {
@@ -116,7 +116,7 @@
           "kind": "record_mass_conservation_receipt",
           "entity": "household",
           "reason": "Amounts-only capital gains redraw: household weights pass through unchanged and total household mass is conserved.",
-          "declared_factor": 1.0,
+          "declared_factor": 1,
           "gate_coupling": "The terminal family gate requires a valid mass-conserving MassChangeRecord carrying exactly this reason."
         },
         {

packages/microcosm-build/src/microcosm/build/uk/country_package.json

@@ -52,6 +52,11 @@
       "kind": "legacy_json",
       "schema_id": "legacy_json"
     },
+    {
+      "path": "frs_release.json",
+      "kind": "legacy_json",
+      "schema_id": "legacy_json"
+    },
     {
       "path": "gates.json",
       "kind": "legacy_json",