HR and WR are all about solving bottlenecks. Inputting and processing 1 atom pr cycle is easy, I have done worse than that many times. I have no doubt I can turn these two inputs into any shape i could imagine stuffing into the output, so the output is what i began by designing. 

Can i construct the output fast enough without a trackloop? Surely not, there is not enough space, too much throughput, and too difficult a bonding pattern. I need a trackloop.
Trackloops are fed with a flat-packed monomer, and extend it into a full monomer on its outputting side. Some experimentation tells me this monomer requires no less than three flatpacked molecules bonded together, complicated by the fact that the only safe bonder placement is the horizontal center bond. This approach got me a loop with potentially 13 r, and i was preparing to solve for this rate.

But then I realized that, while bonding is severely limited, there are two completely legal debonders on the output side. Like you see in this solve; an elbow can be dissasembled to build the entire upper or lower row of the monomer. With this approach, i quickly reached the output loop you see here. It is so perfect, surely Zorflax intended this to be possible. I have found some alternatives that may reach this rate, but none as practical.

After having thrown together a quick monomer production line to verify the solution, came the difficult work of understanding the finer details of monomer construction. This was a lot of trial and error, for which i was not rewarded much. What you see here is only a medium amount of optimizations upon my second viable layout, despite trying several other viable ones after.

I have had a monopoly on this type of trackloop-based HR solve, and monopolies stifle innovation, so I am very interested in competition for 8r. If anyone else gets 8r, i would not be surprised if they beat me in cost.