still wondering if just sampling the threadpool or having each thread register or something like e.g. a

DisplayFuture

that wrapped the threadpools execution of futures (or other shim) wouldn’t be reasonable