BEP-24: Dana and Brian

Abstract:

Random ideas
============
* See if we integrate it with the new Synapses object
* Other option: derive from Connection and overload propagate()
* Group -> NeuronGroup
* in Group: N=(50,100)-> N =50*100 and store the shape of the group somewhere
* Construction of sparse connection matrices with kernels
* Discrete time updates: find a syntax (V(n+1)=f(V(n),...)); update with Euler
  and dt=1.
