researching a way to build a minimal perfect hashmap of words to store alongside the binary note in nostrdb so that nostrdb clients (damus ios, damus notedeck, zaptream rust) can do O(1) mute word matching, which is pretty important for performance when you have 100s of mute words.