Fix TANF asset modeling and immigration timing

[MaxGhenis]

Summary

This PR tightens TANF baseline modeling in three areas:

• populates cited countable-resource formulas for DC, Washington, and Montana using the asset inputs we can defensibly observe in CPS
• fixes monthly TANF eligibility formulas that were reading annual immigration variables on the wrong period
• models DC TANF work noncompliance as a partial cash sanction instead of an all-or-nothing eligibility veto

What changed

• DC TANF now distinguishes applicants from enrolled recipients for work requirements:
  • dc_tanf_eligible requires work compliance for non-enrolled applicants (per DC Code § 4-205.19b), preserving the POWER routing for work-noncompliant applicants
  • enrolled recipients who fail work requirements stay eligible but receive a benefit sanction (per DC Code § 4-205.19f)
  • dc_tanf applies the sanction rate: 6% from 2017-10-01 (30% of the 20% adult portion) and 25% from 2026-10-01 (D.C. Law 26-55)
  • sanction references cite DC Code § 4-205.19f(g) and § 4-205.18(e)
• DC, WA, and MT countable-resource formulas now use cited rules and the best available CPS inputs:
  • liquid assets are modeled directly
  • WA and MT approximate countable vehicle resources using average vehicle value (excludes one vehicle); this is an upper bound when vehicles differ in value — a highest_vehicle_value input would allow the exact rule
  • WA and MT continue to exclude non-homestead real property until we have a defensible input for it
• Monthly TANF eligibility call sites now use period.this_year when reading annual immigration variables across the state TANF formulas touched here, plus Idaho TAFI.

Validation

• policyengine-core test policyengine_us/tests/policy/baseline/gov/states/dc/dhs/ -c policyengine_us (170 tests pass)
• uv run pytest -q policyengine_us/tests/core/test_dc_tanf_work_sanction.py
• uv run ruff check $(git diff --name-only -- '*.py')

Known limitations

• WA/MT vehicle approximation overstates countable resources when vehicles differ in value (e.g., $11k + $1k → $6k countable vs $1k correct). Documented in code comments.
• 11 non-TANF files had the same immigration_status Enum period bug (SNAP, CCAP, AABD, CCFA, etc.) — fixed in Fix immigration_status period access in non-TANF monthly formulas #7990.

Remaining gap work

This PR does not solve the larger TANF eligibility miss in CPS baseline construction. That follow-up is tracked in #7988.

Current worst remaining FY2024 admin-vs-modeled average monthly recipient gaps after this patch:

• Oregon: 0 vs 18,196
• Kansas: 0 vs 2,933
• Maryland: 4,852 vs 12,919
• New York: 97,049 vs 128,335
• New Hampshire: 4,852 vs 2,499

DC is now close enough that it no longer belongs in that blocker set:

• DC: 4,852 vs 5,056

The remaining problem is structural, not takeup priors: TANF application-month income and assistance-unit modeling on CPS still need a deeper pass.

[codecov]

Codecov Report

❌ Patch coverage is 92.91339% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 90.71%. Comparing base (3c2c641) to head (f547037).
⚠️ Report is 21 commits behind head on main.

Files with missing lines	Patch %	Lines
...states/ak/dpa/atap/eligibility/ak_atap_eligible.py	0.00%	1 Missing ⚠️
...lity/ca_tanf_immigration_status_eligible_person.py	0.00%	1 Missing ⚠️
...s_child_care_immigration_status_eligible_person.py	0.00%	1 Missing ⚠️
...lity/il_tanf_immigration_status_eligible_person.py	0.00%	1 Missing ⚠️
..._financial/ma_tafdc_immigration_status_eligible.py	0.00%	1 Missing ⚠️
...states/pa/dhs/tanf/eligibility/pa_tanf_eligible.py	0.00%	1 Missing ⚠️
...ance_unit/tx_tanf_categorically_eligible_person.py	0.00%	1 Missing ⚠️
...tates/wa/dshs/tanf/eligibility/wa_tanf_eligible.py	0.00%	1 Missing ⚠️
...ates/wi/dcf/works/eligibility/wi_works_eligible.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##              main    #7989      +/-   ##
===========================================
- Coverage   100.00%   90.71%   -9.29%     
===========================================
  Files            2       85      +83     
  Lines           33     1152    +1119     
===========================================
+ Hits            33     1045    +1012     
- Misses           0      107     +107     

Flag	Coverage Δ
unittests	90.71% <92.91%> (-9.29%)	⬇️

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.