AI Position of First Cities

The choice of city start positions by AI by  Freeciv 2.0.0 appears 
incorrect.

For example, AI moved a settler from a plains-wheat f/p/t=3/1/0 tile to 
a grassland 2/0/0 tile before founding a city.  If it had founded the 
city on the plains wheat tile, the total production for the city would 
have been plains-wheat 3/1/1 plus forest-pheasant 2/2/0 for a total of 
5/3/1. After the AI move, the total production for the city was 2/0/1 
plus 2/2/0 for a total of 4/2/1, which limits the production of new 
settlers early in the game. This fact means AI play during startup is 
less than optimal.

Below are the results of two games, played twice each (4 game plays), 
that show the current AI position of first cities is less than optimal.

*_Comparative Game Scores:_
*
GAME 1

*Human controls first cities of Player 1 (game 1)*

   Player    Score
        1      311
        2      456
        3      330


*Same game with AI control of Player 1 (game 1 again from 
civgame-4000m.sav.gz)
*

   Player    Score
        2    Destroyed the other two players, on two separate runs of
   the same game in 1876.

GAME 2

*Human controls first cities of Player 1 (game 2)*

   Player    Score
        1      475
        2      370
        3      325


*Same game with AI control of Player 1 (game 2 again** from 
civgame-4000m.sav.gz**)
*

   Player    Score
        1      298
        2      537
        3      276



*
*