Enzyme assisted post-processing isn’t a horrible idea…especially if you are looking to achieve simple things where nature has already provided a solution or model (eg modding rubisco)
We have been “evolving” enzymes for use as bio-catalysts for many years.
…and have recently cracked the code for determining the folding of proteins from their primary sequence (alphafold2)
So picking a problem, then designing an enzyme to solve it, will be the right way to perform a whole heap of things in the not too distant future.
Not convinced that extracting cannabinoids or terpenes from the compartment they were presented to us in is gonna be one of those.