As usual, I build my rate solves single-pipeline.  In this puzzle, though, it might actually be good, since it saves a bunch on duplicate projection glyphs.

Things to note:
1. This solve started with hexarms 1 and 2, which just seemed too nice to build a solution without
2. The solve is 14P, since that avoids parity issues with the two main pulling arms if it were 7P
3. Delivery of the two extra inputs to the output area by hitchhiking on the quicksilver buffer