Finish resolving Refractoriness issues

[thesamovar]

We did have a problem whereby neurons could be refractory and still fire if the synaptic weight was greater than the difference between the reset and threshold values - this is quite a degenerate case but was occasionally cropping up. This is now fixed, but the fix may slow things down a bit, and makes the old approach redundant. This should be fixed. For example, we don't need to clamp state variable values every time step now perhaps, but only at the end of the refractory period (maybe this isn't critical though).

[thesamovar]

It is best to keep the current approach because it is cleaner and changing it requires changing quite a lot of code.