Add 2025 Alabama standard deduction tests for surviving spouse

[jindal-git]

Summary

This is a follow-up to #9424 addressing the post-merge review feedback from @anth-volk:

1. Finding A2 (Surviving Spouse Tests): Added 2025 boundary test cases for SURVIVING_SPOUSE at $25,999 (full $8,500 deduction) and $26,000 (first $175 reduction to $8,325) in al_standard_deduction.yaml.
2. Finding A3 (EOF Blank Line): Removed the trailing blank line at the end of al_standard_deduction.yaml so git diff --check passes cleanly.
3. Finding A1 (Act Citation): Noted that HB 163 was enacted as Act 2022-297 (rather than Act 2022-292).
4. Changelog: Added changelog fragment in changelog.d/.

---

Response to Finding C1 (Rounding Alabama AGI)

We reviewed Finding C1 regarding rounding Alabama AGI prior to floor division and agree that no change should be made to the formula in al_standard_deduction.py:

1. Scope: PR Fix 2025 Alabama standard deduction phase-out threshold #9424 only corrected a parameter regression in threshold.yaml introduced in Update 2025 Alabama income tax #7192. It did not alter the formula logic in al_standard_deduction.py, which has performed floor division on continuous AGI since its initial implementation.
2. Filing Convention vs. Statutory Mandate: The Form 40 instruction ("You may round off cents to whole dollars on your return...") is a taxpayer filing convenience standard across IRS Form 1040 and state returns to simplify paper/e-file line entries, rather than a statutory mandate modifying the legal definition of adjusted gross income for benefit/deduction evaluation.
3. Engine-Wide Consistency & Microsimulation: Throughout PolicyEngine US, continuous float values are preserved throughout intermediate tax calculations rather than rounded to whole dollars at intermediate steps. Introducing localized rounding (e.g. np.round(al_agi)) inside individual deduction variables would create artificial notch/step discontinuities for continuous income distributions in microsimulation and API usage without statutory basis.

For these reasons, the existing unrounded continuous evaluation in al_standard_deduction.py is appropriate and retained.