Add a test that has varying iontype in different sections

[iomaganaris]

During validation of NetPyNE results we figured out that there was an issue with the calculation of iontype for each section. This issue was fixed in #775. We should add a test in the CI that (according to @nrnhines) tests the following:

One would need a model that had time variation of conci
to test that vector.record of erev (or just the final erev on return) from coreneuron psolve was the same as the neuron psolve.