Statistical average activation energy for edges in jump network

[stefsmeets]

Instead of an empirical definition of the free energy from the trajectory, which depends on the quality of the MD simulation, it is possible to implement a more statistical approach. In this approach, the single edge values can be replaced by their statistical average, obtained by averaging similar transitions from the simulation data. It is also possible to add noise and randomicity to this statistical approach and generate random variations of the same graph.

Follows up on #322