examples/income_distribution_us.py does not auto-fetch its dataset

[vahid-ahmadi]

Reported in the discussion on #264.

Unlike the other US/UK examples (e.g. us_budgetary_impact.py uses ensure_datasets() to download from HuggingFace on first run), examples/income_distribution_us.py only checks for a local file at examples/data/enhanced_cps_2024_year_{year}.h5 and raises:

FileNotFoundError: Dataset not found at <path>. Run create_datasets() from policyengine.tax_benefit_models.us first.

(See examples/income_distribution_us.py:29-35.)

This is inconsistent with the rest of the example suite and is a stumbling block for new users (and JOSS reviewers).

Fix: replace load_representative_data() with the same ensure_datasets(...) pattern used in us_budgetary_impact.py so the dataset is fetched on first run.