Reorder blocks in structured trace for more optimal memory

[lucasxia01]

The arithmetic selector is currently active until the end of the aux block. If that holds, we can put the arith and aux gates next to each other in the ordering. The databus inverse polynomials are active until the end of the max of the busread block and the databus data sizes, so it would optimal to put the busread block near the beginning.

Similarly, the lookup inverse polynomial is active over the lookup table region, which is currently at the end. We could consider moving the lookup block to the end if we keep the lookup tables where they are.