invent/implement edge similarity algorithm to efficiently draw edge visualizations

[benmaier]

added the possibility to retrieve an edge similarity matrix in tacoma.get_edge_trajectories

[benmaier]

edge similarity between to edges e = (u,v) and h = (i,j) is defined as follows.

Let's denote a function F(e,t) = 1 if e is in the edge set of the graph at t, 0 else. Then edge similarity is defined as

A_{e,h} = \Theta (-1/2 + \delta_{u,i} + \delta_{v,j} + \delta_{u,j} + \delta_{v,i}) * \int_t0^tmax  dt F(e,t) * F(h,t)