Current state of assembly

[dark64]

Description

At the moment, the usage of assembly has a few implications:

• The output size of the program might be impacted in some cases. This comes from the fact that the witness generation functions generated from the assembly assignment statement are duplicated n number of times in the case of for-loops with a range of 0..n
• Non-optimal performance of assembly blocks in the interpreter (no memoization techniques are implemented yet)

Solution

One of the solutions is to extract solvers to a separate section of the output program and reference them in the IR. This makes it possible to optimize away duplicates and write only what is necessary. A prerequisite to this solution is to define a proper file structure.

These issues are being addressed in #1268

[dark64]

Solved in #1268