Create response areas from each part's answer

[peterbjohnson]

Every ticket writes its tests before its code, and its pull request reports a run over a target set with the number of matches. See docs/plan.md. Write every run's output to a file under out/ before reading it.

The built set has no response areas, so an imported part has no answer box. The ME2 export holds 13 across five questions, of three kinds: NUMERIC_UNITS with comparePhysicalQuantities (pre-text $m=$, answer 0.106 kg), MATH_SINGLE_LINE with symbolicEqual ($F=$, answer (pi/6)*rho*U**2*R**2), and MULTIPLE_CHOICE with arrayEqual (one boolean per option). in2lambda's ResponseArea writes all three.

After the routes have settled a part, one small model call per part receives the part's content, its options and its final answer, and proposes its response areas: kind, pre-text, the answer in the platform's machine form, and the correct options for a multiple-choice part; none for a part that asks for a discussion. The proposal is written through ResponseArea into the set. The measure is the ME2 export: per part, same number of areas, same kind, same answer after normalising units and symbols, and the report counts matches and lists misses. Tests first: tests/test_response_areas.py with a fake backend for the three kinds and the none case, and a live test over the ME2 target. Done when the ME2 pair scores its 13 areas and the misses are in the PR.

---

Workbench ticket t44.

Live run over the ME2 target (manager, 2026-09-23 23:30)

areas     ME2_Fluids_introduction: 11 of 12 match
miss      ME2_Fluids_introduction: q4.p2[2]: wanted NUMERIC_UNITS '0.541 mm', made NUMERIC_UNITS '0.1 mm'

The one miss is a judgement the solutions leave open: the worked solution bounds the particle size between 0.1 mm and 1 mm, and the author's box holds 0.541 mm.