Adding thresholding option to matrix_to_marching_cubes()

[bhacha]

I found that it was a bit of an annoying pre-processing step to convert my numpy arrays to Boolean arrays before using the ops.matrix_to_marching_cubes function. I think adding an optional arg for setting a threshold could be a quality of life improvement.

In this implementation 'threshold' is None by default but if a float or int is passed, that value is used to binarize the input matrix. It converts every Element > Threshold to True and then works the same way as the existing version of the function.

[mikedh]

Awesome thanks for the PR!!

One numpy "gotcha" is that numpy integers and floats aren't always instances of Python integers and floats. I changed the check to threshold is None, or you can also do isinstance(item, (int, np.integer, float, np.floating)). This should be released in #2319

In [3]: a = np.zeros(10, dtype=np.int32)[0]

In [4]: isinstance(a, int)
Out[4]: False

In [5]: isinstance(a, (int, np.integer))
Out[5]: True

In [7]: isinstance(10, (int, np.integer, float, np.floating))
Out [7]: True