Add a next_with_timeout method to DataLinkReciever

[AJMansfield]

One attempt at adding a way to do per-call timeouts on DataLinkReciever.
I added it this way for symmetry with the similarly-named TransportChannel methods, though I'm not sure if this is really the best way to do this; perhaps it would be better to just add a set_timeout method?

[mrmonday]

I think a set_timeout() would be better, yes. Or possibly just an entry in the Config struct?

[AJMansfield]

I think a set_timeout() would be better, yes. Or possibly just an entry in the Config struct?

That does probably fit better with the design of the library; and I'll happily port my application code to use wrappers around that abstraction if it's added upstream instead of using my own fork.

Just to show how I'm using next_with_timeout from my fork, though, I use it specifically for a single-threaded multiplexed eventloop idiom (e.g. the type of eventloop calls like select(2) exist for in C).
With this specific minimal example being to re-transmit packets after a delay:

loop {
    let rx_result = match tx_queue.do_delayed_retransmit(Instant::now()) // also returns tx time for the next packet after
            .map(|t| t.saturating_duration_since(Instant::now())) // how far in the future is that?
        { 
            Some(sleep) => rx.next_with_timeout(sleep), // sleep at most that long, don't miss the next tx time
            None => rx.next(), // sleep as long as possible, there's no next tx
        }.map(|pkt| (pkt, Instant::now())); // TODO: use SO_TIMESTAMP instead to get the packet timestamp from the kernel

    match rx_result {
        Ok((packet, now)) => tx_queue.schedule_delayed_retransmit(packet, now + delay),
        Err(ref err) if err.kind() == std::io::ErrorKind::TimedOut => {},
        Err(e) => panic!("Error recieving packet: {}", e),
    }
}