Add test coverage and enhance documentation for salary sacrifice pensions

- Add comprehensive smoke test for pension_contributions_via_salary_sacrifice
  - Validates non-negative values
  - Checks for presence of data
  - Ensures reasonableness against total employment income
- Enhance inline documentation with detailed explanation:
  - What SPNAMT represents
  - Relationship to other pension variables
  - Rationale for not using outlier clipping (follows job table pattern)
  - Source attribution and data transformation notes

Addresses review feedback on PR #214

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

Author: MaxGhenis

policyengine_uk_data/datasets/frs.py

@@ -631,12 +631,20 @@ def determine_education_level(fted_val, typeed2_val, age_val):
     pe_person["employer_pension_contributions"] = (
         pe_person["employee_pension_contributions"] * 3
     )  # Rough estimate based on aggregates.
+    # Salary sacrifice pension contributions from FRS Job table (SPNAMT field)
+    # SPNAMT represents employer pension contributions made via salary sacrifice
+    # arrangements where employees forego salary in exchange for increased pension
+    # contributions. This is separate from regular employee pension contributions
+    # (deduc1) and provides tax advantages for both employer and employee.
+    # Uses same pattern as employee_pension_contributions (deduc1) without outlier
+    # clipping, as job-level data is generally cleaner than pension provider data.
+    # Source: https://datacatalogue.ukdataservice.ac.uk/datasets/dataset/630d4a8d-ba6a-82b3-f33d-c713c66efcb3
+    # Note: Values are annualized from weekly amounts reported in the survey.
     pe_person["pension_contributions_via_salary_sacrifice"] = np.maximum(
         0,
         sum_to_entity(job.spnamt.fillna(0), job.person_id, person.person_id)
         * WEEKS_IN_YEAR,
     )
-    # link to the variable: https://datacatalogue.ukdataservice.ac.uk/datasets/dataset/630d4a8d-ba6a-82b3-f33d-c713c66efcb3
 
     pe_household["housing_service_charges"] = (
         pd.DataFrame(

policyengine_uk_data/tests/test_pension_contributions_via_salary_sacrifice.py

@@ -0,0 +1,24 @@
+def test_pension_contributions_via_salary_sacrifice(baseline):
+    """Test that pension_contributions_via_salary_sacrifice loads and has reasonable values."""
+    values = baseline.calculate(
+        "pension_contributions_via_salary_sacrifice", period=2025
+    )
+
+    # Basic validation: all values should be non-negative
+    assert (
+        values >= 0
+    ).all(), "Salary sacrifice pension contributions must be non-negative"
+
+    # Should have some non-zero values (not everyone uses salary sacrifice, but some do)
+    total = values.sum()
+    assert (
+        total > 0
+    ), f"Expected some salary sacrifice contributions, got {total}"
+
+    # Reasonableness check: total should be less than total employment income
+    # This is a very loose check just to catch major issues
+    employment_income = baseline.calculate("employment_income", period=2025)
+    total_employment = employment_income.sum()
+    assert (
+        total < total_employment
+    ), f"Salary sacrifice contributions ({total/1e9:.1f}B) cannot exceed total employment income ({total_employment/1e9:.1f}B)"