Implement monte carlo tree search actor

[Klazkin]

Implement an actor that chooses its action based on estimated action values from running simulations and random roll outs. While a lot of the groundwork has already been done, notably in #44. There are still some significant challenges that may occur, notably we must:

• Know which actions can be used.
  • Easy to implement, just use Unit::get_hand()
• Know which combinations can be made from existing actions.
  • Combining an action means that the actions which were used to make the combination can no longer be used.
• Know where those actions can be cast
  • While there may be only a small number of usable actions, ~10, there could be numerous positions where they can be cast. For example, if an actor uses the action that allows them to move to any free position within a radius of 6 cells, this creates around 100 possible new states for that action alone. So instead, I need to implement something that can generate limited set of candidate positions for the actions to be used based on assumptions such as, "an actor rarely want to move in a random direction, most often either towards or away from certain units".

The goal

• [x] Implement generators for possible candidate targets for an action.
• [x] #49
• [x] #48
• [x] Random actor that selects a single random action.
• [x] Random actor that selects random actions.
• [x] Random actor that selects random actions (with action combinations).
• [x] MCTS actor that selects a single optimal action.
• [x] MCTS actor that selects optimal actions.
• [x] MCTS actor that selects optimal actions (with action combinations).

Time tracking

Time Estimate: 3 hours 30 minutes Time spent: 14 hours 30 minutes

Resources

MCTS https://en.wikipedia.org/wiki/Monte_Carlo_tree_search, https://www.geeksforgeeks.org/ml-monte-carlo-tree-search-mcts/, https://webdocs.cs.ualberta.ca/~hayward/396/jem/mcts.html https://www.cs.swarthmore.edu/~mitchell/classes/cs63/f20/reading/mcts.html Time https://www.geeksforgeeks.org/measure-execution-time-function-cpp/

[Klazkin]

for if I forget why this took over 2x times than expected, I still dont know the cause but for some reason my random generator has a 1 in a 100000 chance to generate a number that is outside of the defined range for which it is supposed to generate numbers.

[Klazkin]

There are still some bugs/improvements that can/need to be made, however, due to this already taking way too much time, those improvements will get their own issues. This one is finally closed.