Add SSTB self-employment income variable and split QBID by category

[PavelMakarchuk]

Summary

Closes #7939.

Adds a new sstb_self_employment_income input variable (Person, USD) and refactors the qualified business income deduction so that the §199A(d)(3) phaseout above the §199A(e)(2) threshold reduces only the SSTB component, leaving the non-SSTB component subject to the W-2/UBIA cap as before. This fixes the mixed-filer case (e.g., a doctor who also owns a rental property) where the legacy business_is_sstb boolean forced an all-or-nothing treatment.

Changes

New variables

• sstb_self_employment_income (Person, USD) — SE income from a specified service trade or business under IRC §199A(d)(2). Subject to SECA, treated separately for QBID.
• sstb_self_employment_income_would_be_qualified (Person, bool, default True) — parallels the existing <source>_would_be_qualified flags.
• sstb_qualified_business_income (Person, USD) — SSTB QBI before the §199A(d)(3) phaseout, with QBI deductions pro-rated by gross-income share.

Modified variables

• qualified_business_income — now excludes SSTB SE income from gross QBI and pro-rates the existing QBI deductions by non-SSTB share.
• qbid_amount — computes the §199A(b)(2) per-business limit separately for the non-SSTB and SSTB categories before summing them with the REIT/PTP component. The legacy business_is_sstb flag is preserved for backward compatibility by routing the legacy QBI through the SSTB component when set.
• taxable_self_employment_income, earned_income, market_income, and the IRS gross_income/sources parameter — add sstb_self_employment_income so it flows through SECA, AGI, and downstream income concepts in parallel with self_employment_income.

Notes on the issue's example tests

The integration tests in #7939 use round numbers ($20,000 and $10,000) that assume QBI is not reduced by the SE-tax / health-insurance / pension ALDs. PolicyEngine's existing qualified_business_income formula does subtract those ALDs (per IRS Publication 535 Worksheet 12-A), so the new tests in this PR follow the same approach: I include the issue's two scenarios verbatim but zero-out the QBI ALD inputs so the expected values isolate the SSTB-vs-non-SSTB routing rather than the deduction pro-rating.

Test plan

• policyengine_us/tests/policy/baseline/gov/irs — 725 tests pass
• policyengine_us/tests/policy/baseline/gov/irs/income/taxable_income/deductions/qbid — 34 QBID tests pass (24 existing + 10 new)
• policyengine_us/tests/policy/baseline/gov/irs/self_employment — 11 SECA tests pass (10 existing + 1 new)
• policyengine_us/tests/policy/baseline/gov/states/ca — 520 California tests pass
• policyengine_us/tests/policy/baseline/gov/usda — 384 USDA tests pass
• policyengine_us/tests/policy/baseline/gov/hhs — 386 HHS tests pass
• make format clean

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.09%. Comparing base (7b50e5d) to head (44d3d9d).
⚠️ Report is 85 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff             @@
##              main    #7944      +/-   ##
===========================================
- Coverage   100.00%   99.09%   -0.91%     
===========================================
  Files            2       14      +12     
  Lines           42      222     +180     
  Branches         2        3       +1     
===========================================
+ Hits            42      220     +178     
- Misses           0        2       +2     

Flag	Coverage Δ
unittests	99.09% <100.00%> (-0.91%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[MaxGhenis]

Requesting changes for one blocking regression.

• sstb_self_employment_income is added to person-level market_income, federal gross income, earned income, and SE tax, but not to gov.household.market_income_sources, which drives household_market_income and then household_net_income. Repro on this branch: a one-person situation with sstb_self_employment_income = 100000 returns market_income = 100000, household_market_income = 0, and household_net_income = -23082.287109375 because the income is omitted from household resources while tax liabilities are still included. Please add it to household market income sources, and audit the other parameterized income-source lists that currently count self_employment_income if SSTB income should remain equivalent outside the QBI calculation.

Verified locally: uv run python -m policyengine_core.scripts.policyengine_command test policyengine_us/tests/policy/baseline/gov/irs/income/taxable_income/deductions/qbid policyengine_us/tests/policy/baseline/gov/irs/self_employment/taxable_self_employment_income.yaml policyengine_us/tests/policy/baseline/gov/states/mo/tax/income/deductions/mo_business_income_deduction.yaml policyengine_us/tests/policy/contrib/taxsim/outputs/taxsim_outputs.yaml -c policyengine_us passed: 71 tests.

[MaxGhenis]

Approved after pushing the household market-income fix. Local verification: the household market income fixture passes, and the prior reproduction now returns household_market_income = 100000.