The AI was previously very very very buggy and whilst it sometimes provided results that looked right under the hood not a lot was going right for it.

This is now better. The AI plays a lot more intelligently (although it still struggles in places such as for some reason often avoiding capturing my peices and dropping peices weirdly).

Code need commenting and ensuring it has all of the nessesary destructors etc. but everything is in and working.

Now all classes inherit from a top level class and this top level class has static methods to handle this.

This allows the graphics to update whilst the AI is thinking.

Memory was leaking due to not being able to delete players in the lookahead destructor. Attempted deletion was causing a crash because of vector resizing calling the destructor then the destructor being called again when the object went out of scope. Using shared_ptrs fixed this.

Just about. The AI is not functioning. It take too long to make a decision currently although I intend to thread some of the tree work so hopefully that will speed it up. Additionally it is possible that the number of simulations/lookahead depth is too long. Still, it works.

AI code updated. Should now follow a MCTS Pattern (although currently the result is not used). I think it may be exploring the work result rather than the best however. Will need investigate.

Also having an intermitant assert due to a nullptr in the player bank. No idea what is causing it.

Next job is to start implementing the MCTS part of thing or to have it so that the move found is actually the one used as currently nothing happens.

Code is very buggy and will crash on run

However I am checking it in becuase I need to debug it so preserving it here first.