configurable entrypoints

I'd like to have something like this:

In setup code:

fn main() { 
  // stand-in for real isolation, could set up seccomp, fork off and re-exec ourselves again using e.g. https://github.com/containers/bubblewrap or systemd-run or whatever
  procspawn::init_config("isolated", || libc::setuid(1000));
  // This is the default config
  procspawn::init();

Then, in the client code we can conveniently dispatch into these distinct process configurations:

// A task we want to be isolated; e.g. fetch a URL via HTTP, or perform some computation, or whatever
procspawn::spawn_config("isolated", || reqwest::get("https://example.com"));
// A task which runs with default process privileges
procspawn::spawn(|| std::fs::write("/etc/someconfig.conf", "somevalue"));

Start by reading the existing procspawn::init and procspawn::spawn entry points and the issue's examples for procspawn::init_config and procspawn::spawn_config. Determine how named process configurations should be registered and dispatched, including the proposed isolation setup, then verify that both configured and default tasks work as described.