Calculates global spatial autocorrelation of a raster based on the similarity to
The resulting statistic is such that the more positive the number, the greater
the similarity of values in the raster, and the more negative the number,
the more dissimilar the raster values are.
Since mean and standard deviation are based off of an
Int based Histogram, those values will come from rounded values
of a double typed Tile (FloatConstantNoDataCellType, DoubleConstantNoDataCellType).
This operation requires that the whole raster be passed in;
it does not work over tiles.