[Bug-gnubg] Effective pip count (EPC) bug

[Christopher D. Yep]

Windows XP, GUI version
version 0.14-dev 1.1345 030904
version 0.14-dev3D 1.1345 030904 [bug exists in both old and "3D" versions 
of gnubg]

Set-up the following position:

    GNU Backgammon  Position ID: /wAAAP4BAAAAAA
                    Match ID   : cAkAAAAAAAAA
    +24-23-22-21-20-19------18-17-16-15-14-13-+  O: gnubg
 OO | O                |   |                  |  0 points
 OO | O                |   |                  |
  O | O                |   |                  |
  O | O                |   |                  |
  O | 8                |   |                  |
    |                  |BAR|                  |v (Cube: 1)
  X | 8                |   |                  |
  X | X                |   |                  |
  X | X                |   |                  |
 XX | X                |   |                  |  On roll
 XX | X                |   |                  |  0 points
    +-1--2--3--4--5--6-------7--8--9-10-11-12-+  X: Chris

Then,

show onechequer

gnubg reports that each side has an EPC of 12.578 (and an avg. rolls (to 
bearoff) of 12.578/8.167 = 1.540).  This is incorrect; the correct EPC is 
about 29.

Also, the onechequer statistics reverses the players.  The pips, Avg. 
rolls, Std. dev., EPC, and wastage numbers listed for player "gnubg" are 
actually those of player "Chris" and vice-versa (the above is a symmetric 
position, but change gnubg's position to 7 on the 1 pt. and 1 on the 2 pt. 
for example).

Chris