Move state computation to tensorflow

[Edouard360]

Instead of computing the local states in python, and call sess.run for each square, we should do it on the tensorflow side, with a single call.

Another interesting thing to do would be to keep this strategy only for the border. To quickly analyse whether the point is at the border or not, we would need an operator doing something like:

    [ 0 0 0 0 ]                             [ 0 0 0 0 ]
    [ 0 1 1 1 ]                             [ 0 1 1 1 ]
X = [ 1 1 1 1 ]       to               X' = [ 1 0 0 1 ]
    [ 0 1 1 1 ]                             [ 0 1 0 1 ]
    [ 0 1 1 0 ]                             [ 0 1 1 0 ]