fix: connection pool exhaustion + cross_ref column name

- tenant_db.py: add rollback() before returning conn to pool to prevent
  'idle in transaction (aborted)' state that exhausts the pool
- tenant_db.py: increase pool maxconn from 10 to 20 for better concurrency
- inventory_vehicle_compat.py: fix column name cross_ref_number ->
  cross_reference_number to match actual schema

pos/services/inventory_vehicle_compat.py

@@ -19,7 +19,7 @@ def auto_match_vehicle_compatibility(master_conn, tenant_conn, inventory_id, par
     Searches:
     1. parts.oem_part_number (exact, case-insensitive, spaces stripped)
     2. aftermarket_parts.part_number
-    3. part_cross_references.cross_ref_number
+    3. part_cross_references.cross_reference_number
 
     Returns:
         dict: {'matched': bool, 'matches': [...], 'myes': [...]}
@@ -46,7 +46,7 @@ def auto_match_vehicle_compatibility(master_conn, tenant_conn, inventory_id, par
     if not oem_ids:
         cur.execute("""
             SELECT DISTINCT part_id FROM part_cross_references
-            WHERE REPLACE(UPPER(cross_ref_number), ' ', '') = %s
+            WHERE REPLACE(UPPER(cross_reference_number), ' ', '') = %s
         """, (clean_pn,))
         oem_ids = [r[0] for r in cur.fetchall() if r[0]]