When there are sure mines, consider them in model evaluation

[gdahia]

Whenever a sure mine is found, it is not currently accounted for in subsequent model evaluations. This would be an improvement over the number of possible models evaluated and moves considered.

[gdahia]

The interpretation for this pheonemon is: opening the board might make a previously sure mine be ambiguous, meaning that not only information might be lost with cell openings, but a different sequent of moves might not shed light on the same mines.

[gdahia]

ie, do not count flagged cells as closed.

[gdahia]

it is not as simple as not counting flagged cells as closed.

[gdahia]

possible implementation by @pedrovidal is fixing the flag value in model evaluation as 1. therefore, we could skip models not satisfying this assingment by doing an and between a bitstring with this info and the model and seeing that it is different from the original bitstring

[pedrovidal]

closed in branch flag