Fast three way join algorithm

Hi there, I just watched this talk https://youtu.be/FudEqfutW2A?t=1888 from MSFP by @henglein and @mkm about how to use module theory to get a very fast implementation of three way joins which probably has more implications with regards to database performance (with a haskell implementation which I think isn't public yet).

In the q&a section there are questions about backporting this approach to RDMS, and generally make this result more visible to the database community, and I thought this might be an idea that would be a great fit for `project-m36`.

Hope this is interesting 😄