FractionalVolume with surface_fraction=0 does not work for 3d Region

[rgourdine]

Example:

dend1 = h.Section("dend1")
dend1.nseg = 4

cyt = rxd.Region(dend1.wholetree(), "i", geometry=rxd.FractionalVolume(volume_fraction=0.8), dx=0.25)
ca = rxd.Species(cyt, name="ca", charge=2, initial=1e-12)
dend1.L = 10
dend1.diam = 2

rxd.set_solve_type(domain=[dend1], dimension=3)
h.finitialize()
print(sum(ca.nodes(cyt).volume)

Expected behavior:
Not raising an exception

[adamjhn]

I think the reason we have not encountered this error before is in cyt the surface_fraction=0 but the nrn_region='i'. So at initialization we try to calculate how to scale the currents. The error in species.py occurs because surface_areas1d return 0 rather than a list.
@ramcdougal Should we throw an exception if surface_fraction=0 and nrn_region='i', or just add warning and avoid adding currents to that region?

[ramcdougal]

The reason this wasn't an issue for the calcium wave experiments was because the FractionalVolume region with no surface area was not in the i region. Furthermore, when we did have it, there was an i region that did have surface area.

Physically this makes no sense... you can't be directly adjacent to the plasma membrane (the i region) and have a total area of zero.

Given that it is convenient to declare things to be in the i region as this allows their concentrations to interact with NMODL and be readable from HOC, I'm inclined to say let's do the warning case where we avoid adding currents... but if so, we should also raise a warning in 1D.

We should also go back and figure out why we made this test, and make sure that the resulting test has a physically meaningful i region and does not raise this warning.