Trivial improvment of cost

Update: I'm probably settled with this, so it's time to write a detailed explanation.

First, I liked the idea of LexC, mainly skilled in area optimization, I'm pretty afraid in attempting min cycle solve before for its complexity. With LexC as the metric, I could first come up with a solve with min latency ignoring the rate, which is exactly what I did this week. 

I'm pretty confident with the primary I get, unless I missed something very important. The bottleneck of latnecy in this week's output is the three air atoms in the product. After grabbing a non-air atom at glyph, it need to be Cal then Dup< then Bond(1/2/3) to turn into product. Dup< must happen after Cal, and Bond could happen anywhere in between. 

To minimize latency, Bonding of each non-air atom can only happen once, which is 3 Ops in total. This naturally leads to the plan: put the air atom at 2-bond position, whose 3 Ops would be Bond twice and Dup> once. The 3-bond air would be bonded by tribond, and the another Dup> could be provided by the wheel. 

To summerize, in 3 step, each atom grabbed would go through:
1-bond Air: Bond Cal Dup<
2-bond Air: Bond Dup> TriBond
3-bond Air: Cal Dup< TriBond 

The remaining problem is how to order the Ops to make them possible. The Dup> can't take the first position, TriBond requires complex maneuver so Bond must happen in T+1 (Grab is T+0). This put the core constraint on the layout, 2 of 4 atoms must be move to bonder in 1 cycle, and 1 of the arms must have a path to complete remaining Ops without regrab. 

My first attempt is putting TriBond in T+3, which works but the arm position conflicts with track loop so I could only get a rate 6 solution. Then I tried to put TriBond in T+2 so in next move the atoms could be moved at a whole with only 1 arm. I tried two unsuccessful layouts, where air-air-air-salt could be make, but the path for second Quintessance keeps crashing into my arms. This was solved in the third layout, which is the one I submitted here. Also I found the sliding output in this solve at first try, I'm not sure but it seems outputing with any other arm would only increase the area.