Deduplicate and Refactor optimization code about neuronunit HOT 4 OPEN

I m not exactly sure, but I think I may have addressed moving duplicate rheobase search code from nsga.py -> exhaustive_search.py. The greater problem is how to package it up as a Neuronunit test object inside tests/init.py still remains.

from neuronunit.

Hi @rgerkin, when we had the meeting in person the Friday a week and a day ago. You mentioned an approach to putting the Rheobase method back inside the tests/init.py collection of neuronunit test classes (in a way that from SCOOPs perspective, it would be an un-nested function (level 0). What was that approach again? I feel like that will be the ultimate way to de-duplicate code should it work.

Also in that meeting I believe you said that the error should be derived from the 'sort_key', as the sort key is a way of normalizing different score metrics into one inter-operable score scheme.

from neuronunit.

Move a lot of the stuff that you use to test your code into a scratch directory outside of the neuronunit module (but inside the neuronunit directory itself, i.e. at the same level a setup.py, unit_test, etc.

from neuronunit.

For now we just need to do enough of this to push out 0.2, i.e. all if it needs to be outside the core of neuronunit, in the optimization directory, and extra cruft needs to be removed.

from neuronunit.