Call for possibly-under-documented Stan tricks for benchmarking project

Hey folks! @kn2465 has taken on a Google Summer of Code project where they will be collecting a variety of models for and seeking to create compute-optimized versions for inclusion in posteriorDB and benchmarking generally. While the Stan User Guide already includes many great optimization suggestions, I thought it would be prudent to also poll the community for any tricks folks might have developed on their own that are not already in the SUG. I’ve started a Google Doc here with some of my own, so feel free to add yours there if you have any!