UK compute campaign, run 4: produce the first populace-uk reform_validation.json (activates the #65 UK lane)

The country-generalised reform-validation ingest (#65) ships with a deliberately empty UK release map: it activates on the first real reform_validation.json from a managed populace-uk run, and fails loudly until then. This issue produces that artifact.

Steps

• Define the UK battery from what runs 1–3 (#76–#78) verify: every UK comparison row whose status lands ok/match_* at registration time (mode-1 counterparts, reckoner deltas, OBR costings), each with its executed baseline key per the #71 registry — the US artifact shape, GBP, populace-uk-rv- run prefix.
• Producer script on the managed side emitting reform_validation.json per (pe-uk release × certified bundle), engine pins from the release manifest — mirroring the US producer's contract documented in ingest_reform_validation.py.
• First artifact lands in sources/populace-reform-validation/raw_uk/, its release id + pin into COUNTRIES["UK"]["engine_versions"], ingest, and the app's per-release history view gets its first UK release.
• Regression semantics from day one: a row leaving tolerance on a later pe-uk release is a first-class event, not a re-snapshot.

This is what makes the UK scorecard continuous — each pe-uk release (and each fiscal event's parameter update) gets a scored regression record. Depends on #65 merging and #76 running.