When I first encountered Tarcles as a metric, I had no idea what I was doing.  Since I've not seen it even once between then and now, I still can't say have much experience to draw on.  But I have significantly leveled-up my general competence as a solver, so I did manage to formulate a plan of attack.  Here's what I was thinking.

*********

Sum and TI are two other metrics that generally reward low arm counts.  And one concept that recurs often in those spaces is identifying *repetition* within the products so that you can reuse machinery.  It wasn't super obvious to me in the first daunting impression, but this bulky product contains quite a lot of repetition--it is the same 3-atom elbow, bonded 4 times around a Nerve Block input.  Once I noticed that, I figured I would try to make those elbows as efficiently as possible and proceed from there.

It seemed to me that best number of arms for making that elbow would probably be 2. The central challenge is getting the mors and vitae separated by a knight's move, and accomplishing that requires either two mechanisms or a lot of regrabs and shuffling. The hexarm here does most of the work, assisted by a simple arm moving mors from debonder bonder.  

The other critical concept I did not originally understand when working on that old Tarcles puzzle was using multiarms to perform multiple tasks and/or minimize reset cycles.  That manifested here by upgrading Arm 2 to a triarm.  That allows the machine to move finished elbows out of the hexarm's workspace without needing multiple costly reset instructions.  Aditionally, this upgrade also allowed Arm 2 to be the output arm, which saved a modest amount of otherwise-empty tape.

************

Taken together, we have the solve you see here.  Hexarm 1 bonding 2/3 of the elbow atoms and setting up the necessary knight's move.  Triarm 2 finishing each elbow and swinging them clear to the product assembly station.  Piston arm 3 on a much slower cadence, pulling the core Nerve Block and manipulating it so that all the elbow bonds line up properly.  

I quite like this solve.  It has a grand total of 3 empty spaces in the tape and very few unnecessary movements.  I can image there might be solves that create elbows more quickly, and thus do better on the metric even though they have more arms. But this is plenty good enough for me.