Optimize the Transition Systme Lowering

We can probably simplify the construction of the ITEs a bit. Right now the longest example is unsigned_mul which takes ~1m in release mode after I did some simple optimization tricks. It takes > 3m in Debug mode

Locate the transition-system lowering implementation and the unsigned_mul example mentioned in the issue. Measure its construction time in release and Debug modes, then inspect how the ITEs are built and identify equivalent simplifications. Done means preserving the lowering behavior while reducing the reported runtime.

Written by the indexing model from the issue text.