I'd be more inclined to be worried about whether the neurons on one side can usefully bind to the neurons on the other. AFAIK the nerve mapping in the spinal chord is more or less random. The rat experiment wouldn't test this, of course.
[ * ] Disclaimer: while I have in-depth and extensive knowledge of some subjects, this is not one of them.
Maybe it would be possible to make a electronic, prosthetic section of spine that can be reconfigured on the fly, a bit like an FPGA, that might be useful for both injuries (where a damaged section needs to be replaced) and clean cuts with unknown mapping. Unfortunately these kinds of brain-computer interfaces seem prone to scarring, weakening the signal over time.
Ah, here's a short article on the work with rats: http://www.iflscience.com/brain/new-drug-boosts-nerve-regrow...