todo

- optparse
- endgames
- evaluation with count of cells to move
- memory leak (garbage collect board)


- analyze + analyze deeper with first 2-3 moves
- openings tests
- tests more
- tests for moves_to_consider
- test on sorting capture_moves
- test on repetitions
- format_move with double pieces

- 3 -> 4
- tests on max_deep_capture, ...
- preanalyze and then analyze deep with sorting
- tests on evaluation

- analyze 2 best moves then analyze them deeper
- analyze one move fast and then with evaluation other moves
- len(pieces) -> total evaluation
- better positions eval
- fast checks


- integer evaluation first
- dp
- two moves at once and then parallel
- better evaluation function but 3-2->3-1
- rook positional eval
- real evaluation function

- if no captures due to check
- kings evaluation from deep=3 forced checks


- material
- space
- activeness
- center
- kings position


- fix bug with board.evaluation break
- rooks connection with mask
- increase deep regarding to len(board_moves)