Smoothed histogram methods more consistent with Python

Smoothed NetEMD calculations now use the same variance calculation as the Python code. For our set of random test networks, the R and Python smoothed NetEMDs are almost identical for graphlet orbits up to 4 nodes (to at least 4 decimal places for all but one pair).

Note: R and Python smoothed NetEMDs are still significantly different for graphlet orbits up to 5 nodes. Investigation of this difference continues.

Fast NetEMD implementation using graphlet orbit degree distributions

Fast NetEMD implementation using Graphlet-based Degree Distributions for all orbits in graphlets comprising up to 4 or 5 nodes. Execution can be parallelised across multiple threads on a single machine and execution time is within a factor of two of the existing Python implementation.

Note: NetEMD for smoothed graphlet orbit counts is not yet working correctly.

Bugfix for issue #9

Fixed issue #9: Filename pattern ignored when reading all graphs from folder

Initial internal test release

Internal test release of initial basic NetEMD functionality