Call reset kernel only with as many threads as there are spiking neurons (not as there are neurons in total)

[denisalevi]

I just saw that we are calling the reset kernel with as many threads as neurons in the population, but only as many threads as there were spikes in the current time step.

Instead, just call the reset kernel with as many threads as there are spikes. For that we need the number of spikes. Which we already need for other kernel invocations and which we should first implement to be copies to host after each thresholder using asynchronous memcpy (see #146).

Once that is implemented, also close #149.