feat(data): Phase 2b complete — EU NUTS-2 spatial join + US state income

- stg_regional_income: expanded NUTS-1+2 (LENGTH IN 3,4), nuts_code rename, nuts_level
- stg_nuts2_boundaries: new — ST_Read GISCO GeoJSON, bbox columns for spatial pre-filter
- stg_income_usa: new — Census ACS state-level income staging model
- dim_locations: spatial join replaces admin1_to_nuts1 VALUES CTE; us_income CTE with
  PPS normalisation (income/80610×30000); income cascade: NUTS-2→NUTS-1→US state→country
- init_landing_seeds: compress=False for ST_Read files; gisco GeoJSON + census income seeds
- CHANGELOG + PROJECT.md updated

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

scripts/init_landing_seeds.py

@@ -15,14 +15,22 @@ import json
 from pathlib import Path
 
 
-def create_seed(dest: Path, content: bytes) -> None:
-    """Write content to a gzip file atomically. Skips if the file already exists."""
+def create_seed(dest: Path, content: bytes, *, compress: bool = True) -> None:
+    """Write content to a seed file atomically. Skips if the file already exists.
+
+    compress=True (default) writes gzipped content, suitable for all landing zone
+    files. compress=False writes raw bytes — required for files consumed by DuckDB
+    ST_Read (e.g. GeoJSON), which cannot read .gz files.
+    """
     if dest.exists():
         return
     dest.parent.mkdir(parents=True, exist_ok=True)
     tmp = dest.with_suffix(dest.suffix + ".tmp")
-    with gzip.open(tmp, "wb") as f:
-        f.write(content)
+    if compress:
+        with gzip.open(tmp, "wb") as f:
+            f.write(content)
+    else:
+        tmp.write_bytes(content)
     tmp.rename(dest)
     print(f"  created: {dest}")
 
@@ -87,6 +95,8 @@ def main() -> None:
             json.dumps({"rows": [], "count": 0}).encode(),
         "eurostat/1970/01/nama_10r_2hhinc.json.gz":
             json.dumps({"rows": [], "count": 0}).encode(),
+        "census_usa/1970/01/acs5_state_income.json.gz":
+            json.dumps({"rows": [], "count": 0}).encode(),
         "eurostat_city_labels/1970/01/cities_codelist.json.gz":
             json.dumps({"rows": [], "count": 0}).encode(),
 
@@ -97,9 +107,21 @@ def main() -> None:
             json.dumps({"rows": [], "count": 0}).encode(),
     }
 
+    # Uncompressed seeds — required for files consumed by ST_Read (cannot read .gz)
+    uncompressed_seeds = {
+        # Empty NUTS-2 boundary placeholder so stg_nuts2_boundaries can run before
+        # the real file is downloaded via scripts/download_gisco_nuts.py.
+        # ST_Read on an empty FeatureCollection returns 0 rows (graceful degradation:
+        # all locations fall back to country-level income until the real file lands).
+        "gisco/1970/01/nuts2_boundaries.geojson":
+            b'{"type":"FeatureCollection","features":[]}',
+    }
+
     print(f"Initialising landing seeds in: {base}")
     for rel_path, content in seeds.items():
         create_seed(base / rel_path, content)
+    for rel_path, content in uncompressed_seeds.items():
+        create_seed(base / rel_path, content, compress=False)
     print("Done.")