Port the NHS budget and the NHS per-capita cost by age and sex behind microcosm's ETB services anchors so the last hand-copied E6 anchors can retire

Microcosm's UK etb_services stage still carries two hand-authored anchor resources flagged chronicle_candidate: true (uk/etb_services_anchors.json, uk/nhs_consumption_by_age_gender.json). The other E6 anchor files retire in microcosm#890's spine half now that their facts are vendored from the 474a0ae artifact (VEH1103, DESNZ/HMRC road fuel, NEED, Ofgem cap levels); these two are the last hand-copied publisher numbers on that surface, and the doctrine (chronicle#221 / #230) is that publisher facts reach Microcosm only as hash-pinned Chronicle artifacts.

Facts to port

1. NHS England budget, FY2025-26: nhs_budget_2025_26 = 202,000,000,000 GBP, cited to HM Treasury, Spending Review June 2025 (E03349913_HMT_Spending_Review_June_2025_TEXT_PRINT.pdf). Microcosm allocates the ETB survey's NHS-in-kind benefit to persons by age and sex and scales to this budget. Ask: a facts-only package for the NHS England resource (RDEL) budget by fiscal year from the Spending Review 2025 tables (FY2023-24 outturn through the settlement years), entity: government, period_type: fiscal_year, with the table and line named in the record set, so the consumer can bind FY2024-25 (the base year) as well as FY2025-26.
2. NHS activity/cost per capita by age band and sex: nhs_consumption_by_age_gender.json (rows by half-open age band × sex, carried from the incumbent uk-data build, cited only as "incumbent public NHS age-gender activity/cost table"). Ask: identify and package the publisher table this came from (the candidates are the ONS "Effects of taxes and benefits on household income" methodology's per-capita NHS allocation by age and sex, or the DHSC/NHS England programme budgeting per-capita costs by age band) as a facts package with dimensions age_band (lower/upper bounds explicit) and sex, entity: person, unit GBP per person per year, and the year stated. If Chronicle cannot identify a public publisher artifact for the incumbent table, say so on the issue and Microcosm will keep the resource with a documented provenance gap instead of a chronicle_candidate flag.

Not asked

• rail_fare_index_2023 in the same file is a PolicyEngine UK parameter path (gov.dft.rail.fare_index) kept for engine lockstep, not a publisher fact; the ORR fares index it is checked against is already in the artifact (orr.rail.fare_index, chronicle #254). It stays.

Context

• Consumer: microcosm packages/microcosm-build/src/microcosm/build/uk_runtime/etb_services.py (etb_services_configuration, allocate_nhs_by_age_gender), stage declaration in uk/spec/sources.yaml (allocate_per_capita_from_cell_table, budget_resource: etb_services_anchors.json).
• Once packaged, Microcosm's #890 spine half (or a follow-up) vendors the rows with tools/vendor_uk_ledger_facts.py, retires the two JSON files and moves the uk_aggregate_admin NHS anchor (nhs_spending_total, 202bn, tolerance 30.3bn) onto the vendored value.

Start with the two resources named in the issue and the consumer entry points packages/microcosm-build/src/microcosm/build/uk_runtime/etb_services.py and uk/spec/sources.yaml. Identify the publisher tables and provenance for the NHS budget and age-gender cost rows, including explicit dimensions, units, and years, then package the facts and document any provenance gap. Done means Microcosm can vendor both resources through tools/vendor_uk_ledger_facts.py and retire the hand-authored anchors.