as we move more work into the engine (i.e. all `Task`s), I’d expect the amount of post-fork work to shear away until we’re effectively able to just call the engine and directly exec an external entrypoint from the thin client runner. so yeah, sort of - but an arduous path to get there fwict.