fix: remove CSV fallback from theme catalog generation, add Parquet debug step

- Remove CSV fallback logic (Parquet-only in M4 migration)
- Add better error messages when Parquet file missing or empty
- Add workflow debug step to inspect Parquet file after tagging
- Simplify build_theme_catalog function signature

.github/workflows/build-similarity-cache.yml

@@ -88,13 +88,60 @@ jobs:
             echo "ERROR: Tagging completion flag not found"
             exit 1
           fi
+      
+      - name: Debug - Inspect Parquet file after tagging
+        if: steps.check_cache.outputs.needs_build == 'true'
+        run: |
+          python -c "
+          import pandas as pd
+          from code.path_util import get_processed_cards_path
           
-          # Verify theme catalog was generated
+          parquet_path = get_processed_cards_path()
+          print(f'Reading Parquet file: {parquet_path}')
+          print(f'File exists: {parquet_path.exists()}')
+          
+          if not parquet_path.exists():
+              raise FileNotFoundError(f'Parquet file not found: {parquet_path}')
+          
+          df = pd.read_parquet(parquet_path)
+          print(f'Loaded {len(df)} rows from Parquet file')
+          print(f'Columns: {list(df.columns)}')
+          print('')
+          
+          # Show first 10 rows with their themeTags
+          print('First 10 cards with themeTags:')
+          print('=' * 80)
+          for idx, row in df.head(10).iterrows():
+              name = row.get('name', 'UNKNOWN')
+              tags = row.get('themeTags', [])
+              tag_count = len(tags) if isinstance(tags, list) else 0
+              print(f'{idx}: {name}')
+              print(f'   Type: {type(tags).__name__}')
+              print(f'   Count: {tag_count}')
+              if tag_count > 0:
+                  # Show first 5 tags
+                  sample = tags[:5] if tag_count > 5 else tags
+                  print(f'   Tags: {sample}')
+                  if tag_count > 5:
+                      print(f'   ... and {tag_count - 5} more')
+              else:
+                  print(f'   Tags: (empty)')
+              print('')
+          "
+      
+      - name: Generate theme catalog
+        if: steps.check_cache.outputs.needs_build == 'true'
+        run: |
           if [ ! -f "config/themes/theme_catalog.csv" ]; then
-            echo "WARNING: Theme catalog not found, generating..."
+            echo "Theme catalog not found, generating..."
             python -m code.scripts.generate_theme_catalog
+          else
+            echo "Theme catalog already exists, skipping generation"
           fi
-          
+      
+      - name: Verify theme catalog and tag statistics
+        if: steps.check_cache.outputs.needs_build == 'true'
+        run: |
           # Detailed check of what tags were actually written
           python -c "
           import pandas as pd