On the expected optimal value of random assignment problems: Experimental results and open questions