Set is_blind from the FRS blind registration flag (SPCREG1) #523

Summary

The FRS records whether an adult is registered blind or partially sighted with their local authority or social services:

• SPCREG1: "Whether registered blind with LA/SS".
• SPCREG2: "Whether LA/SS registered partially sight".

The FRS build never sets PolicyEngine UK's is_blind, so it is False for everyone in the Enhanced FRS.

In FRS 2024-25 (raw adult table, household grossing weight gross4), 29 adult records (57.3k weighted) have SPCREG1=1 and 59 records (123.1k) have SPCREG2=1.

Where is_blind is read in policyengine-uk (main)

• Pension Credit severe disability addition: the blind-partner route (SPC Regs 2002 Sch I para 1(1)(c)), and whether a blind resident's presence is ignored (para 2(2)(b), #1951).
• Council Tax Reduction non-dependant exemption.
• TV licence concession.
• Disabled Students' Allowance qualifying condition.
• Childcare child-age rules: Tax-Free Childcare, the WTC childcare element and HB childcare.

Open questions before mapping

1. Registration vs certification. The legal tests say "certified as severely sight impaired or blind by a consultant ophthalmologist" (for example SPC Sch I para 2(2)(b)). Is local-authority registration an accepted proxy? Check the DMG and the certification and registration process in each nation.
2. Partially sighted. Should SPCREG2 (partially sighted, or sight impaired) map to anything? It does not meet the "severely sight impaired" tests.
3. Children. Check whether the child table carries an equivalent.

Proposed change

Set is_blind = (SPCREG1 == 1) in frs.py. Add a differential test showing that the weighted count matches the raw table, then measure the Enhanced FRS impact on each consumer listed above in a real run.

Found while checking PolicyEngine/policyengine-uk#1951: PolicyEngine/policyengine-uk#1951 (comment). No counted resident in #1951's affected groups is registered blind, so this does not change that PR's numbers.