Run the uk-childcare-100k lane on policyengine.py 6.11.1's certified pair #232

Why

The uk-childcare-100k lane's latest certified result runs policyengine.py 6.11.0's pair (#228). policyengine.py 6.11.1, the latest release (uploaded 2026-10-27), certifies a later pair:

• national cut microcosm-uk-2024-25-national-20261025T091414Z-a6299504
• policyengine-uk 2.151.0

On the 6.11.0 pair, PolicyEngine's static cost exceeds CenTax's by 26% in 2027-28 and 25% in 2029-30. Universal hours add £0.143bn and £0.159bn of that. The whole universal change falls on children whose family holds extended take-up true and universal take-up false. microcosm#1338, merged 23 October, nests the universal draw under the extended one. This cut was built on 25 October.

Done means

• Add the pair to data/uk/certified_pairs.json, with the same identity checks as Add a certified pair registry so the £100k childcare lane can run on policyengine.py 6.11.0's Microcosm pair #229.
• Confirm the cut carries the nested draw, and measure whether any universal change remains.
• Run the compute script with --variant <pair> for 2026–2031 under the same identity checks, with the full-use diagnostic years.
• Re-check every engine- or run-based note at 2.151.0, add the pair to variants, and write replacements where a note no longer holds.
• Add a verified_pairs entry with a pe_gap decision.
• Update the README's comparison table and gap section. History keeps the 19 June and 6.11.0 runs and the published row.

The pair still lacks CenTax's calibration targets, so a result here doesn't settle the dashboard's republication.