days 4-6.83: identifying control flow constructs was way, way, way harder than expected. i'm finally getting somewhere with the analysis. but i'm not even close to using this data to improve the &asm-trace interface.