Add Arch-backed UK target provider

[MaxGhenis]

Summary

• add a UK Arch consumer-fact JSONL target provider
• map OBR receipt/expenditure concepts, HMRC salary-sacrifice employee-count and relief facts/projections, HMRC SPI income-band targets, ISC private-school pupil counts, SLC student-loan/support targets, ONS/NRS demographic counts, ONS savings interest, ONS dwelling-tenure and household-type targets, Scottish Child Payment spend, regional council-tax band counts, DWP claimant/benefit-cap/UC payment-distribution/two-child-limit targets, NTS household vehicle-availability targets, and the EHS/PRHI housing rent targets into canonical microplex target specs
• preserve Arch source lineage, geography, concepts, source filters, source names, and component lineage for repayment, benefit-cap, and projection targets in target metadata
• normalize UK country and region GSS codes to PolicyEngine enum values and support PE-specific fiscal-year labeling where needed
• apply PE target-profile construction choices explicitly where they are not source facts, including the 1.9 property-income scale on HMRC SPI amount targets, the 3% annual HMRC salary-sacrifice relief uprating from the 2024 base, the NRS live-birth proxy for Scotland babies-under-1, the DWP benefit-cap monthly-band midpoint annualization/flooring, the NTS share-times-household-base conversion, the EHS weekly-social-rent-times-52-times-UK-social-renter-base conversion, the PRHI monthly-private-rent-times-12-times-UK-private-renter-base conversion, and the ONS National Balance Sheet 2024 total-land rounding/2020 sector-share split/hold-forward profile while preserving source details in target metadata
• map ONS household-type rows to PE household-type count profiles, including child-count, head-age, household-person-count, and household-benefit-unit filters where PE uses those profile definitions
• map DWP UC monthly payment-band facts to PE annual UC filters and family-type filters; the open-ended top band preserves PE name parity with the current nan_to_nan label while retaining a truthful lower-bound filter
• map DWP two-child-limit publication rows selectively to the 15 PE 2026 target profiles, leaving Arch-only unsupported breakdown rows unmapped
• support tax-year/fiscal-year/monthly/calendar-year consumer facts such as 2022-23, 2023-24, 2025-05, 2025-11, 2026, and 2024 NTS shares and align the Python floor/lockfile with the shared microplex dependency
• verifies Arch local-authority council-tax facts map to PE voa/council_tax/{LA}/{band} target names and local_authority filters without changing production mapping code

Verification

• uv run --extra dev ruff check src/microplex_uk/targets/arch.py tests/targets/test_arch.py
• uv run --extra dev pytest tests/targets/test_arch.py -q -> 24 passed
• uv run --extra dev --with tables pytest -q -> 49 passed
• real HMRC salary-sacrifice relief 2024 suite smoke check: projected all 6 PE HMRC relief targets for 2025 and 2026 with matching names and values under PE's 3% annual growth rule
• real SLC borrower forecast 2026 suite smoke check: built 4 valid consumer facts and matched all 4 PE 2026 borrower target names and values
• real EHS suite smoke check: built 1 valid consumer fact and matched PE housing/rent_social for 2025 name, source, unit, variable, and value (£30.68bn)
• real ONS PRHI private-rent 2026 suite smoke check: built 1 valid consumer fact and matched PE housing/rent_private for 2025 name, source, unit, variable, and value (£89.0352bn)
• real ONS Families and Households 2024 suite smoke check: built 10 valid consumer facts and matched all 10 PE 2024 household-type target names and values
• real ONS National Balance Sheet 2024 suite smoke check: built 5 valid consumer facts and matched PE ons/household_land_value, ons/corporate_land_value, and ons/land_value for 2024, 2025, and 2026
• refreshed PE parity gap count with land suite included: 2025 covered 278 of 3,157 PE targets; 2026 covered 28 of 975 PE targets
• real VOA + Scottish Government council-tax 2025 combined suite smoke check: 2,662 / 2,662 PE voa/council_tax target names covered with 0 missing names and 0 value mismatches, including 2,563 local-authority band targets
• uv run --extra dev ruff check src/microplex_uk/targets/arch.py tests/targets/test_arch.py
• uv run --extra dev pytest tests/targets/test_arch.py -q -> 24 passed
• uv run --extra dev --with tables pytest -q -> 49 passed

Source smoke checks

• built Arch OBR expenditure 2025 suite: 19 valid consumer facts
• built Arch HMRC salary-sacrifice reform 2025 suite: 3 valid consumer facts
• built Arch HMRC salary-sacrifice relief 2024 suite: 6 valid consumer facts mapped to 6 PE 2024 relief targets and projected to all 6 PE 2025/2026 relief targets
• built Arch HMRC SPI income-band 2023 suite: 156 valid consumer facts mapped to 156 2023 income-band targets with total-income filters and positive-income count filters
• built Arch ONS UK population projection 2025 suite: 13 valid consumer facts mapped to UK total and gender-age targets
• built Arch ONS Families and Households 2024 suite: 10 valid household-type count facts mapped to all 10 PE 2024 ONS household-type target names with matching values
• built Arch ONS National Balance Sheet 2024 suite: 5 valid land-value facts; the total-economy fact maps to PE national total, household, and corporate land targets using explicit rounding, split, and hold-forward metadata
• built Arch NRS mid-year population 2024 and vital-events 2024 suites: Scotland under-16 and babies-under-1 proxy targets mapped
• built Arch ONS dwelling-tenure 2024 suite: 5 valid consumer facts mapped to 5 England tenure targets, with social rent represented as the PE council-or-HA tenure union
• built Arch MHCLG EHS rented-sectors 2024 suite: 1 valid £118/week social-rent fact mapped to PE housing/rent_social with source-to-target adjustment metadata
• built Arch ONS PRHI private-rent 2026 suite: 1 valid £1,374/month UK private-rent fact mapped to PE housing/rent_private with source-to-target adjustment metadata
• built Arch ISC census 2024 suite: 1 valid private-school pupil-count consumer fact with concept alignment
• built Arch SLC borrower forecast 2025 and 2026 suites, student-support 2025 suite, and repayment suites for 2025
• built Arch ONS savings-interest 2025 suite: 1 valid consumer fact
• built Scottish Government social-security-assistance 2025 suite: 1 Scottish Child Payment spend fact mapped, with the separate under-1 child-count fact intentionally not mapped here
• built VOA and Scottish Government council-tax-band 2025 suites: 2,662 PE voa/council_tax target names matched exactly, including the 2,563 source-numeric local-authority band facts from VOA CTSOP1.0
• built Arch DWP PIP daily-living, benefit-statistics, and benefit-cap 2025 suites: 29 valid consumer facts mapped to 8 2025 DWP targets, including ESA/JSA counts and benefit-cap total reduction
• built Arch DWP UC national payment distribution 2025 suite: 104 valid consumer facts mapped to all 104 PE UC payment-distribution target names
• built Arch DWP UC two-child-limit 2026 suite: 42 valid consumer facts, selectively mapped to all 15 PE two-child-limit target names with matching values and no extra target names
• built Arch DfT NTS household car availability 2024 suite: 3 valid share facts mapped to all 3 PE NTS household vehicle-count targets with matching values and source-to-target adjustment metadata
• provider smoke-loaded 30 targets for 2024 from HMRC pension tax relief, ISC census, ONS tenure, ONS household-type, ONS land, NRS demographic, and NTS consumer facts
• provider smoke-loaded 286 targets for 2025 across 27 Arch source suites, including HMRC relief projections, ONS land targets, EHS/PRHI rent targets, 104 UC payment-distribution targets, and 18 SLC targets with repayment plan rollups
• provider smoke-loaded 28 targets for 2026, including HMRC relief projections, ONS land targets, SLC borrower forecasts, and all 15 PE 2026 DWP two-child-limit targets; no 2026 SLC target names remain missing in that year profile