Selective Deepening Generators

Since we will be looking at a subset of all possible moves, we need to create a method which prunes all legal moves to moves that have the qualifications we are looking for.

Thus, I am expecting that we create some method something along the lines
`generate_qualifying_moves()`
which will return a list of candidate moves we are allowing our alpha beta pruning to operate under.

Note: this should be done in a specific order of precedence. According to a paper, the following should be the order (for these generators):

1. Capturing pieces that the opponent just played
2. A Killer move (???) [Paper on Killer Moves](https://dl.acm.org/doi/epdf/10.1145/800179.810240)
2. Responding to opponent attacking moves (prevent attacks from coming to fruition)
3. Attacking moves
4. Defensive moves

Start by locating the existing alpha-beta pruning and legal-move generation entry points, then determine how a generate_qualifying_moves() method would fit between them. Done means returning candidate moves in the stated precedence order: captures, killer moves, responses to attacks, attacks, and defensive moves.

Written by the indexing model from the issue text.