Clustering - don't build shadow matrix for a single condition

[GoogleCodeExporter]

The cluster system in Biodiverse allows users to specify one or more sets of 
spatial constraints, resulting in one or more dissimilarity matrices.  The 
system then uses a shadow matrix to keep track of all the changes across both 
matrices to ensure correct calculations of linkages when the first matrix is 
exhausted and clustering shifts to the next matrix.

Of course, if there is only one matrix then there is no need for a shadow 
matrix, and having it needlessly uses memory and processing time.

Original issue reported on code.google.com by shawnlaffan on 21 Jan 2011 at 2:48