Opus Magnum keeps separate leaderboards for cheapest price, fewest cycles used, and smallest area. I generally tended to optimize for fewest cycles and ended up with the kinds of stuff you describe for Codex.
If you open up a least-symbol solution for Codex you'll find a lot of interesting tricks used to reduce symbols. For example, grab-left-left-release-left-left can be accomplished with just grab-left-release-wait, if you sync it so the released atom is removed by another arm at exactly the right time. It's a nice feeling to successfully shave off even one more symbol using these hacks!
In Opus Magnum, there are a lot of interesting things you can do to get more parallelism going in your solution. Except that, again, "fake" difficulty is added in the form of restricted inputs. Maybe I could build the molecule faster with two sources of water, but I only have one, and one source can't produce more than one particle per two cycles. The parts with costs are unlimited -- they just count against your cost. Source and sink parts are "free", but limited in bizarre ways. So much for the parallel approach I wanted to try. :/