L=9 if grabbing the input by a salt-like; L=10 if grabbing by a silver. the L=9 approach needs quicksilver from a seventh input to complete the final product in 22C. 

rather than spend all week failing to achieve this, i decided to go for the straightforward min+1 strategy where every input is grabbed by the silver. 

spare latency in the odd-input pipeline is used for cost savings. critelli is unused. probably possible to get much better secondary at 23C with a more area-conscious design.