Investigate issues related to resetting the network

[rowleya]

When running the "AsyncNeuromorph" code, a network reset didn't seem to work correctly. Rough network description:

• Each "layer" has three populations, two excitatory with 250 and 750 neurons respectively, and one inhibitory with 250 neurons.
• Each of the populations is fed with a Poisson source on a one-to-one basis (like normal ;)
• The neuron model is IFCurrDeltaCa2Adaptive

There are potentially 169 layers (it is actually a 13x13 grid), though I can't see why that would matter.

After the reset and before the run, the i_alpha parameter of the neurons of the excitatory populations are set, and the rate of each Poisson source is also set (so it goes set_params -> run -> reset -> set_params -> run).

My current guess is that it is the model that might not quite be right in setting parameters after the reset, or else the Poisson source doesn't reset correctly after the rate has been set.

Everything works perfectly if you do setup -> set_params -> run -> end -> setup -> set_params -> run -> end.

[rowleya]

I guess a simple test is to do single IFCurrDeltaCa2Adaptive pop fed by Poisson population, and then try setting things and see if it does the same when doing reset or setup... end.