Inputs should not have to include PGN spikes

[MJC598]

Part of the generated data is PAG spikes, but MPG spikes should work sufficiently well given the biological example. This transition cannot happen easily though as the generated PAG_spikes.csv file is utilized throughout the project. This should be generalized to work for either PAG or MPG.

[MJC598]

The feedback loop requires a low level neuron and a set of afferents. This means that generate_input.py and build_network.py will need to be adjusted as will the simulation_config.json. They all depend on a PAG afferent rate due to node set: PAG_aff_virt. Perhaps generalizing this in the generate_input.py would make this work?