* Split informant in two * Clean up imports * Don't rely on saturated_into * Add some documentation to the informant display