Identify cyclic frayed ropes?

[fedarko]

If we allow cyclic bubbles (#241), then I guess we should also allow cyclic frayed ropes for the sake of consistency? I don't think this will mess with "interpretation" of the frayed rope pattern, since the structure still converges then diverges (see Miller et al. 2010).

Allowing this should be fairly simple, since it's just a matter of removing an explicit check I added in (shown below):

    for n in ending_node_ids:
        # Check for extraneous incoming edges
        if len(g.pred[n]) != 1:
            return ValidationResults()
        for o in list(g.adj[n].keys()):
            # We know now that all of the ending nodes only have one
            # incoming node, but we don't know that about the starting
            # nodes. Make sure that this frayed rope isn't cyclical.
            if o in starting_node_ids:
                return ValidationResults()

Update: extra motivation to leave this in (and something to look for once it's merged in): there are some obvious frayed ropes that aren't tagged currently because of cyclic edges like this - for example in the august1 srs049959 graph:

Unless I'm missing something, I think this update should allow these ropes to be tagged.