Fix tax_unit_itemizes tied-federal-tax fallback (#8210)

[PavelMakarchuk]

Closes #8210. Supersedes #8209.

Bug

When tax_liability_if_itemizing == tax_liability_if_not_itemizing (common at low/moderate incomes where refundable EITC + ACTC absorb any federal-tax difference between the two branches), tax_unit_itemizes falls back to:

state_standard_deduction < state_itemized_deductions

These helpers have asymmetric state coverage:

• state_standard_deduction.py lists 8 states (CT, ID, MO, ND, NM, SC, UT, CO) that adopt the federal standard deduction.
• state_itemized_deductions.py lists only 5 states (CT, GA, ND, SC, UT) for federal itemized. For CO/ID/MO/NM/VT/LA/etc., it sums state-specific line items, which for many states is $0.

For a CO joint filer with $50k of itemizable expenses and tied federal tax: state_standard_deduction = $31,500, state_itemized_deductions = $0, fallback returns False → don't itemize. TaxAct/TAXSIM both itemize.

Downstream effects

State	TAXSIM ticket	Mechanism
CO	#870	CO taxable income = federal taxable income; standard-fed gives wrong CO tax
VT	#869	vt_cdcc = 0.72 × capped_cdcc; capped_cdcc depends on federal tax → cascades
LA	#854	Same mechanism via capped_cdcc
NJ #865, DE #850, etc.	various	Likely same root cause; pending triage

Fix

When federal tax is tied, fall back to the federal-level deduction comparison (standard_deduction vs itemized_taxable_income_deductions). This matches:

• The non-branching path already in this file.
• The IRS / TaxAct / TAXSIM-comparison-mode default of itemizing whenever itemized > standard at face value.

Daniel Feenberg confirmed in the issue that TAXSIM's PE-comparison mode uses this naive rule.

Note on the helper variables

state_standard_deduction and state_itemized_deductions remain in the codebase. They map to TAXSIM v34 ("Standard Deduction") and v35 ("Itemized Deductions") outputs in policyengine-taxsim's variable_mappings.yaml. tax_unit_itemizes simply no longer reads them internally.

Tests

• 7 unit tests in tax_unit_itemizes.yaml — tied-tax cases retargeted to federal-level inputs.
• me_deductions.yaml — tied-tax case retargeted (same input pattern).
• CO integration (new): 2025 joint filer with $30k property tax + $20k mortgage, asserts tax_unit_itemizes: true. Locks in TAXSIM MA income tax docs are broken #870 fix.
• VT integration (new): 2025 joint filer with property tax + mortgage, asserts tax_unit_itemizes: true and vt_cdcc: 0. Locks in TAXSIM Create issues for each program x state combination with the GitHub API #869 fix.
• All 13,825 state-level tests pass; 817 federal IRS tests pass.

Residual gaps to TaxAct (out of scope)

• CO income tax is ~$65 short of TaxAct's -$4,397 (CO add-back issue).
• VT income tax is ~$1 short of -$3,045 (rounding).

Both are pre-existing state-model issues — the fix gets us from "very wrong" to "within rounding/state-add-back precision."

Test plan

• make format clean
• All directly-affected tests pass locally
• Full state-level test suite passes (13,825 / 13,826 → 13,826 after ME fix)
• Full IRS test suite passes (817)
• CI green

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (d55cd4f) to head (9c44328).
⚠️ Report is 22 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main     #8213   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            2         1    -1     
  Lines           35        18   -17     
  Branches         0         1    +1     
=========================================
- Hits            35        18   -17     

Flag	Coverage Δ
unittests	100.00% <100.00%> (ø)

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.