so this might be the god-solve

OPTIONAL EXPLANATION IF THIS IS A GOOD SOLVE:
The crux of the problem is to figure out how to divide the reagent in half and insert it into the same processing pipeline. This can be done, quite cleverly, by placing the halves on two adjacent corners of an imaginary hexagon, rotationally symmetric about the hexagon's center, and using a bi-arm placed at the center to pick up one at a time, one per arm. I previously had a four arm solution doing just this.

The true aha moment comes when one realizes that we don't even need to place the halves to be rotationally symmetric about the imaginary hexagon's center; the center arm can use any number of pivots to compensate for the difference in rotation. This allowed me to reduce the number of arms to three.