`build_compartment_graph` doesn't follow min_radius when computing electrical properties

[bantin]

Hi Jaxley team -- congrats on the paper! Really nice to see that it's finally out.

Anyway, I noticed an issue with the way that build_compartment_graph handles the min_radius argument. Right now, the morph_attrs_from_xyzr uses the values of the radii before they are clipped to the value of min_radius. This means that effectively the min_radius argument is ignored when setting the properties of the cell.

I noticed this when I was trying to load a (weird) SWC file where the radii were all zeros. I thought that I would be able to get a working model by passing min_radius, but the resulting values like resitive_load_in, resistive_load_out were all NaN.

Suggested fix: Before calling swc_radius, swc_area, swc_volume, or swc_resistive_load in morph_attrs_from_xyzr, clip the xyzr radii to min_radius (when provided). I'm happy to submit a PR for this if helpful.

[michaeldeistler]

Hi Ben, thanks for reporting! @jnsbck could you have a look while refactoring the SWC reader?

[jnsbck]

Will do! :)