[Bug-gnubg] zp with new weights performance

[Joseph Heled]

mgnu_zp (0ply moves and cube) has played 815 matches with the new 
weights. Its projected rating is 2010. As usual, if you take only 
matches with opponents rated higher than 1700, it's rating is 2045. If 
you take only opponents lower, it is 1910. This is in line with my 
calculations (and others observations) that the FIBS formula is skewed 
when rating difference is too large.

On the whole it is a nice result. Seems not very far from mloner and 
mloner_II, which played recently too (but on plys 2 and 3!)

I wish I had a fast computer which could host the 2ply version 
(mgnutest) at a reasonable speed. Any linux with a processor above 1G 
will do. If you know someone who might want to help (even for a limited 
time) let me know.

-Joseph