"How much metal shavings can we add to Kazyans sandwich before he notices?"

This puzzle would be brutal to optimize, which is why I didn't. I did my best to spot shaves as I was programming it, and did fully program out both viable layouts, but I'm not going to watch this cycle by cycle to make sure it's perfect. I'm sure this solve bleeds cycles in places, but I'm simply too lazy. I'm very curious to see the commentary for this metric. I'm either missing some of the solution space, or it's going to be very difficult to tell where solves save cycles