Implement basic performance benchmark

[TheButlah]

Continuation of the discussion in #77.

We need to implement a basic performance benchmark, to detect any major performance regressions. docs.rs/criterion seems like a good choice. However, Its proving a lot more difficult than expected to use criterion to benchmark async code - its not clear to me exactly how to do this - I can't for example pass the &mut Criterion through to the async task, because its not Send

[TheButlah]

Yep, getting all manner of issues because criterion wants to pass a non-static ref along to a non-async function, so to invoke an async function I would need to to task::spawn() which needs a static ref. If anyone has any experience using criterion in async code, let me know

[TheButlah]

Update: I think I've figured out how to get this working. Here's what the code looks like right now. The key was manually creating the tokio runtime, and using block_on to block until the future completes. I think that this is the right way to do things:

fn criterion_benchmark(c: &mut Criterion) {
    let mut rt = Runtime::new().unwrap();
    let (mut pubs, mut subs) = rt.block_on(setup(4, 16));

    c.bench_function("m pubs and n subs messaging", |b| {
        b.iter(|| rt.block_on(f(&mut pubs, &mut subs)))
    });

    async fn f(pubs: &mut [PubSocket], subs: &mut [SubSocket]) {
        // Send/Recv messages
    }
}

criterion_group!(benches, criterion_benchmark);
criterion_main!(benches);