Distance Restraints for the general non-linear cases

[fgrunewald]

Why?

Originally posted by @pckroon in https://github.com/marrink-lab/polyply_1.0/pull/151#discussion_r716620237

[pckroon]

Ah ok, I understand the algorithm better now. Generating the same bfs traversal in two places will be super fragile. How about the following:

1. generate restraints along the shortest path between the relevant nodes. Also annotate those with the predecessor(s?) along that path.
2. When doing the BFS (so a different path) to create the coordinates, check if the node has any annotated restraints. If so, honour them, if not, continue.

I think the trick here is to annotate not just the distance, but also relative to which preceding node. Or even just annotate the first node relative to which the dis_res needs to be.