FRS 2024-25 refresh of the microcosm-built spine

[juaristi22]

Workstream E11 of the UK-migration master epic #665; child of #145. Explicitly after the swap-acceptance child signs off — 2023 is the parity vintage because the frozen incumbent reference is enhanced_frs_2023_24.

Scope

• Re-pin the stage-manifest artifacts to the FRS 2024-25 release (survey year 2024, calibration year 2025) and the matching non-FRS vintages; re-run the scale ladder.
• Declare period alignment everywhere a fact year differs from the survey year; coordinate fact vintages with WS-B (wave-2 facts) and WS-F (uprating, Decide UK panel, public transfer, and other product-scope leftovers from uk-data #148 — calibration-year materialization is WS-F machinery this child consumes, not implements).
• Adopt the incumbent's release-object discipline (one frozen release object naming survey/base/calibration years, study number, DOI, zip sha) as a declarative resource; label tape years by tax year (e.g. 2024_25) per the SPI-stage convention already adopted from policyengine-uk-v2 lessons.
• Retire hmrc_income_source_stages.json and its lockstep test (the frozen June evidence anchor is superseded by the refreshed pins).

Acceptance

• The build materializes at the refreshed vintage end to end; every artifact pin updated; the battery passes at the new vintage.
• No period alignment left implicit.

Depends on: E10; WS-F for calibration-year materialization.