Modified the old PreferentialAttachment to correctly use the
random generator of GeneratorBase (there was a bug if one changed
the random seed, it was not taken into account).

Also added a Barabàsi-Albert generator (BarabasiAlbertGenerator ...)
that does exaclty the same job but can produce graphs, not only trees.
Indeed the generator adds one node at each call to nextEvents(), for
this node it can add one edge (chosen randomly, with a bias toward
node with high degrees). However doing so creates a tree. One can
now make two ore more edges for each node added. The number of edges
generated at each node can be set. It can also be chosen randomly
(uniformly this time).

The BarabasiAlbertGenerator is stricly the same as the
PreferentialAttachmentGenerator when the number of edges added per
node is one. We can however keep the old generator for compatibility
reasons, or replace it by the BarabasiAlbertGenerator by changing
the name of one of the two.

Also added the citation of the Barabàsi-Albert model used by
this generator.

Also added the start of a generator framework to create graphs from
Delaunay/Dirichlet/Voronoi tessellations.

Fix a bug creating an infinite loop and add some code to ensure that the produced graph is connected.

Add feature to handle an non-connected graphs. In such case, result will be a spanning trees forest.

BaseGenerator adds the hability to backup some data on which nodes
and edges are created. However when creating a second graph with the
same generator, this data must be cleaned up. This has been added
in a default "end()" method. Each generator has been modified to
call this super.end() method.

The algorithm is now largely tested in weighted and unweighted
graphs and should work on non-multi-graphs. It works on directed
graphs, but ignore the direction.

Added the possibility to register a progress "callback" since
the algorithm can take a lot of time to compute.

Also removed the extraneous @override in Community.

Merge branch 'feature/community-measures' of http://github.com/gjherbiet/gs-algo into gjherbiet-feature/community-measures