grafx - Extensible Java Graphics Package Code

Made some computer player tests and decided that the strategy
QBEST_STRAT_WEIGHTED_2_BEST_FUTURE_SCORES is better than
QBEST_STRAT_OPTIMAL_FUTURE_SCORE especially for extremely difficult levels,
therefore sticking with that strategy.

Incentivizing an empty well eval score, by bypassing the next eval after that.

Removing TODO.txt comment related to giving bonus eval score for three or more
layers cleared, because that's already implemented.

Fixing spelling error in comment in ffmpeg-overlay.sh.

Implementing short circuit method for end of level, for qBestStrategy other
than QBEST_STRAT_OPTIMAL_FUTURE_SCORE.

Fixing a bug where end of game was not being seen one move ahead of time,
because the wrong layersToClear value was being passed to evalWithNextBlock().
qSearchBreadth is being reverted to 24, since the recent change to 32 had
nothing to do with fixing this problem. Also adding a TODO.txt item related to
picking low-hanging fruit related to the situation where a pit/well is cleared
of all cruft, which usually results in a bonus points situation.

In ComputerPlayerLevelAction changing qSearchBreadth from 24 to 32. I found
that near the end of level it wasn't always seeing the end sequence.

Trying to make the multi clear layers bonus incentive more configurable. Now
we should be able to set the layer count to 2 instead of 3.

Implemented the three layers cleared bonus strategy but I'm not so sure that it
works, yet.

Making incremental changes to incentivize the clearing of three or more layers
at once. For now added new field in inner class BlockPlacementStrategy.Move,
named 'threePlusLayersCleared'. Reduced SUPERMAXIMAL_EVAL_SCORE to 10.4
because we're expecting to define THREE_LAYERS_CLEARED_SCORE_BOOST as 0.3, and
the sum of the two will be 10.7, which is still allowable.