a simple in-memory database, trying to copy the best solutions from datascript,
xtdb, fulcro, autonormal and especially shadow-grove. each of them has its own
strengths, and is an excellent solution, but each lacks some of the cool things
that the others have. doxa tries to take the best from every solution while
being damn fast and lightweight.