Add detailed skip tracking for phone imports

- Track skipped_no_phone and skipped_no_id separately
- Display skip information in admin UI with warning icon
- Clarify that empty phone numbers cannot be imported (PK constraint)
- Update documentation to explain expected skip behavior
- Example: 143 rows without phone numbers is correct, not an error

When importing PHONE.csv with empty phone numbers:
- Rows are properly skipped (cannot have NULL in primary key)
- User sees: '⚠️ Skipped: 143 rows without phone number'
- This is expected behavior, not a bug

app/import_legacy.py

@@ -773,7 +773,7 @@ def import_rolodex(db: Session, file_path: str) -> Dict[str, Any]:
 
 def import_phone(db: Session, file_path: str) -> Dict[str, Any]:
     """Import PHONE.csv → LegacyPhone model with upsert logic."""
-    result = {'success': 0, 'errors': [], 'total_rows': 0, 'updated': 0, 'inserted': 0, 'skipped': 0}
+    result = {'success': 0, 'errors': [], 'total_rows': 0, 'updated': 0, 'inserted': 0, 'skipped': 0, 'skipped_no_phone': 0, 'skipped_no_id': 0}
     
     try:
         f, encoding = open_text_with_fallbacks(file_path)
@@ -789,7 +789,12 @@ def import_phone(db: Session, file_path: str) -> Dict[str, Any]:
                 rolodex_id = clean_string(row.get('Id'))
                 phone = clean_string(row.get('Phone'))
                 
-                if not rolodex_id or not phone:
+                # Skip rows with missing required fields (phone is part of PK, cannot be NULL)
+                if not rolodex_id:
+                    result['skipped_no_id'] += 1
+                    continue
+                if not phone:
+                    result['skipped_no_phone'] += 1
                     continue
                 
                 # Create a composite key for tracking