Use Populace geoslices in policyengine.py

[MaxGhenis]

Summary

• make populace_us_2024 the default US dataset used by create_datasets, load_datasets, and ensure_datasets
• scope US state and congressional district regions from the certified national Populace H5 via state_fips and congressional_district_geoid, including the at-large SS00 district convention
• stop vendoring derived Populace state/CD H5 slices into the runtime bundle manifest; keep places as hierarchy metadata until Populace emits place_fips
• harden dataset reference resolution so bundled logical names win over same-named local files, while explicit local paths still require unmanaged opt-in for managed resolution
• regenerate bundle TRO sidecars with pe:emittedIn=repository-bundle

Validation

• uv run --python 3.13 ruff format ... && uv run --python 3.13 ruff check ...
• uv run --python 3.13 --extra us --extra uk --with pytest python -m pytest tests/test_us_regions.py tests/test_release_manifests.py tests/test_entity_utils.py
• uv run --python 3.13 --extra us --extra uk --with pytest python -m pytest tests/test_scoping_strategy.py tests/test_entity_utils.py tests/test_us_regions.py tests/test_release_manifests.py tests/test_certify_data_release.py tests/test_bundle.py tests/test_bundle_metadata.py tests/test_dataset_sources.py tests/test_region.py tests/test_trace_tro.py
• HUGGING_FACE_TOKEN=... HF_TOKEN=... uv run --python 3.13 python scripts/bundle.py check --include-tros
• git diff --check

Notes

A local full-CD Populace build was interrupted by the computer crash before release manifest completion, so this PR does not certify or publish any new Populace dataset. It only updates policyengine.py to consume the already-certified national Populace artifact as the default and derive geoslices by row filtering.