This also removes some cruft, and makes handling of {CXX,LD}FLAGS more
consistent.

This also fixes a potential problem with layered SBMs, where the move
probabilities would be incorrectly computed.

This changes the underlying adjacency matrix implementation to a single
mixed in + out edge list for each node, instead of separate ones. This
simplifies the undirected_adaptor<> so that its out-edge iterators
become trivial, and also faster.

This fixes issue #362.