Differential Item Functioning in Computerized Adaptive Testing: Can CAT Self-Adjust Enough?