grafx - Extensible Java Graphics Package Code

Reverting back to the old linear combination of metrics. There is some magic,
some secret sauce which makes this combination work extremely well. I simply
cannot explain it, other than stating the observation that this combination
seems to consistently pass the hardest impossible level, and every single time
at that.

Trying some new metrics. These are not working very well.

Adding upperLeftSectionLowerMetric(). Defining linear scalars for two new4
metrics.

A first pass at plugTheHoleMetric(). Untested.

Making variables 'final' where I can, and now using the new
elevationAtGridLocation() function where I can. No logic/algorithm changes.

In EvalFunction, defining elevationAtGridLocation(), and using it in
overallHeightMetric() and in heightFluctuationMetric(). Meanwhile I found a
blaring problem in overallHeightMetric(), which had to do with the computation
for maxAltitudesPossible, which used the z-dimension twice instead of using
the y-dimension. This has been fixed, and the linear factor
OVERALL_HEIGHT_METRIC has been changed to compensate for the correction.

Nightmare level default, changed from index 4 to index 2.

This eval function does very well on the hardest level, level 23, even with the
adjustments made, causing it to have an extremely difficult block distribution.
I tested this and the computer player passed this level at least 5 times in a
row, with not a single failure, with these eval function settings.

Adding comment for plugTheHoleMetric(). This needs some very careful
consideration.

Beginning to work on plugTheHoleMetric().