Fine-tuning conductance models

[jgostick]

Extracted networks occasionally have issues with pores that overlap too much (see #2363). Too much means the center of one pore is beyond the lens of the other pore.

There are two solutions we should implement:

1. Add 2 new conductance models: Intersecting cones and intersecting pyramids (see #1960). These models require a 'throat.centroid' which we currently output in snow2 as throat.global_peak. So the models would have an argument like throat_centroid='throat.global_peak'. We should probably export throat.centroid from snow2 in the near future.
2. Update the existing models to throw a hard exception if pores overlap too much...instead of trying to do an 'opinionated' correct on the fly.

[jgostick]

Option 2 is probably a good idea with or without option.

Notes:

• It might be nice to do this as a health check when you actually run a simulation, BUT the overall conduit conductance values may be positive even in any of the individual elements had a negative conductance. Also, it's is possible that the conductance being used is NOT coming from a size factor, so checking them would stop simulations that it shouldn't. Basically, there is no general way to check this at run time, so we have to add the checks to the size factor models directly.