Computes histograms for n-dimensional data (i.e. compute
number of data points in a n-cube).

The code is designed to be fast, and optionally uses a c-mex file for
the core routine. The c-mex file is not necessary and you get
good performance even without it.

The c-file must be stored under the correct name and compiled
with mex for optimal performance.

Hans.Olsson@dna.lth.se