Lukas Drescher

Thank you @wvangeit for the review! The tests should pass now. Please let me know if there are open issues. I'll have to prepare an example comparing Arbor with NEURON...

> Ok, thank you, that already looks nice. One thing I think would be useful is to have a test that actually runs arbor on the generated code and compares...

In f2ca1ed, I've added support for inhomogeneous parameters, which was the last missing feature of the cable cell exporter that I'm aware of. > One thing I think would be...

In e68599075fa, I've added a functional test based on the deviation of voltage traces in the L1-norm between Arbor and Neuron (relative error < 5% with `dt = 0.025` gives...

I've uploaded a set of validation runs [here](https://github.com/lukasgd/BluePyOpt-utils/tree/master/validation/all-regions) and added some [analysis](https://github.com/lukasgd/BluePyOpt-utils/blob/master/validation/all-regions/arbor_neuron_validation_analysis.ipynb).

So, maybe I should add a bit of explanation regarding the validation runs and analysis. The filename schema for the [output notebooks](https://github.com/lukasgd/BluePyOpt-utils/tree/master/validation/all-regions) is `l5pc_soma_arbor_[__...].ipynb`, where `` describes the L5PC section...

All requests should be addressed, `prune_tag` and `median_distal_radii` are removed (see comments for details), `split_at` fixed, tests and documentation added.

Changed the `split_at` implementation again to avoid redundant loops on large models. I've changed the node predicate as well to a segment id predicate - the node struct contains the...

Thanks for the feedback. @brenthuisman, I've resolved all points I've addressed in 622f204. Feel free to reopen any if you'd like further changes.

I've swapped the label-dict and decor arguments to the `cable_cell` constructor and in the returned values of `read_acc`. @brenthuisman, is there anything more from your review that needs to be...