Using the 'merge parallel edges' functionality

[Robinlovelace]

Looking at the docs, nx_merge_parallel_edges seems to be just what we're looking for in the 'networkmerge' project, or at least a part of the solution.

I plan to give it a go, just checking-in, are there any considerations when working with large datasets? We're likely to run this for all of Scottish road network if we get it working and wondering if it will work better if we regionalise, doing it for each of 33 local authorities. Thanks for great documentation by the way, could be v. helpful for our work!