fix: performance improvements, shared UI, and cross-reference data quality

Backend (server.py):
- Fix N+1 query in /api/diagrams/<id>/parts with batch cross-ref query
- Add LIMIT safety nets to 15 endpoints (50-5000 per data type)
- Add pagination to /api/vehicles, /api/model-year-engine, /api/vehicles/<id>/parts, /api/admin/export
- Optimize search_vehicles() EXISTS subquery to JOIN
- Restrict static route to /static/* subdir (security fix)
- Add detailed=true support to /api/brands and /api/models

Frontend:
- Extract shared CSS into shared.css (variables, reset, buttons, forms, scrollbar)
- Create shared nav.js component (logo + navigation links, auto-highlights)
- Update all 4 HTML pages to use shared CSS and nav
- Update JS to handle paginated API responses

Data quality:
- Fix cross-reference source field: map 72K records from catalog names to actual brands
- Fix aftermarket_parts manufacturer_id: correct 8K records with wrong brand attribution
- Delete 98MB backup file, orphan records, and garbage cross-references
- Add import scripts for DAR, FRAM, WIX, MOOG, Cartek catalogs

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

vehicle_database/scripts/create_cross_references.py

@@ -0,0 +1,125 @@
+#!/usr/bin/env python3
+"""
+GENERADOR DE REFERENCIAS CRUZADAS ENTRE MARCAS
+Encuentra partes de diferentes fabricantes que cubren los mismos vehículos
+y crea referencias cruzadas bidireccionales entre ellas.
+"""
+
+import sqlite3
+from pathlib import Path
+from collections import defaultdict
+
+DB_PATH = Path(__file__).parent.parent / 'vehicle_database.db'
+
+
+def get_db():
+    conn = sqlite3.connect(DB_PATH)
+    conn.row_factory = sqlite3.Row
+    return conn
+
+
+def main():
+    print("=" * 70)
+    print("GENERADOR DE REFERENCIAS CRUZADAS ENTRE MARCAS")
+    print("=" * 70)
+
+    conn = get_db()
+    cursor = conn.cursor()
+
+    # Get existing cross-ref count
+    cursor.execute("SELECT COUNT(*) FROM part_cross_references")
+    existing_xrefs = cursor.fetchone()[0]
+    print(f"\nCross-refs existentes: {existing_xrefs:,}")
+
+    # Step 1: For each part_group, find parts from different brands
+    # that fit the same vehicle (model_year_engine)
+    print("\n[1/3] Buscando partes que cubren los mismos vehículos...")
+
+    # Build a map: (group_id, mye_id) -> list of (part_id, part_number)
+    cursor.execute("""
+        SELECT vp.model_year_engine_id, vp.part_id, p.oem_part_number, p.group_id
+        FROM vehicle_parts vp
+        JOIN parts p ON vp.part_id = p.id
+        WHERE p.group_id IS NOT NULL
+        ORDER BY p.group_id, vp.model_year_engine_id
+    """)
+
+    group_mye_parts = defaultdict(set)
+    for row in cursor.fetchall():
+        key = (row['group_id'], row['model_year_engine_id'])
+        group_mye_parts[key].add((row['part_id'], row['oem_part_number']))
+
+    print(f"      Combinaciones grupo+vehículo: {len(group_mye_parts):,}")
+
+    # Step 2: For each (group, vehicle) with multiple parts from different brands,
+    # create cross-references
+    print("\n[2/3] Generando pares de cross-reference...")
+
+    # Build existing cross-ref set for fast lookup
+    cursor.execute("SELECT part_id, cross_reference_number FROM part_cross_references")
+    existing = set()
+    for row in cursor.fetchall():
+        existing.add((row['part_id'], row['cross_reference_number']))
+
+    print(f"      Cross-refs existentes en set: {len(existing):,}")
+
+    # Collect new cross-reference pairs
+    new_xrefs = []
+    for key, parts_set in group_mye_parts.items():
+        if len(parts_set) < 2:
+            continue
+
+        parts_list = list(parts_set)
+        for i in range(len(parts_list)):
+            pid_a, pn_a = parts_list[i]
+            for j in range(i + 1, len(parts_list)):
+                pid_b, pn_b = parts_list[j]
+
+                # Skip if same part number prefix (same brand)
+                if pn_a[:3] == pn_b[:3]:
+                    continue
+
+                # Add A->B
+                if (pid_a, pn_b) not in existing:
+                    new_xrefs.append((pid_a, pn_b))
+                    existing.add((pid_a, pn_b))
+
+                # Add B->A
+                if (pid_b, pn_a) not in existing:
+                    new_xrefs.append((pid_b, pn_a))
+                    existing.add((pid_b, pn_a))
+
+    print(f"      Nuevas cross-refs a crear: {len(new_xrefs):,}")
+
+    # Step 3: Insert
+    print("\n[3/3] Insertando cross-references...")
+    inserted = 0
+    for i, (part_id, xref_number) in enumerate(new_xrefs):
+        if i % 5000 == 0 and i > 0:
+            print(f"      Insertando {i}/{len(new_xrefs)}...")
+        cursor.execute(
+            "INSERT INTO part_cross_references (part_id, cross_reference_number, reference_type, source) VALUES (?, ?, 'interchange', 'Vehicle Fitment Match')",
+            (part_id, xref_number))
+        inserted += 1
+
+    conn.commit()
+
+    # Final stats
+    cursor.execute("SELECT COUNT(*) FROM part_cross_references")
+    total_xrefs = cursor.fetchone()[0]
+
+    conn.close()
+
+    print("\n" + "=" * 70)
+    print("CROSS-REFERENCES COMPLETADAS")
+    print("=" * 70)
+    print(f"""
+RESUMEN:
+  - Cross-refs antes:      {existing_xrefs:,}
+  - Nuevas cross-refs:     {inserted:,}
+  - Total cross-refs:      {total_xrefs:,}
+""")
+
+
+if __name__ == '__main__':
+    main()