[Bug-gnubg] Fix for ratings in relational database

[Jim Segrave]

The calls to PythonGame had the ismatch and match-length arguments
reversed with respect to the function, so for matches, all rating
calculations were done on one point matches. Now fixed, but for those
who have been using it and care, you'll need to either manually fix or
repopulate the database.

I must say, there are some intersting things one can see in the
database by selecting various fields and then plotting them. 

Does anyone think adding tables for games within a match/session would
be useful? If the table gamestats existed, then you can do a join to
get a list of all the games in a match, or examine how you handle
Crawford or post-Crawford games.

-- 
Jim Segrave           address@hidden