Race condition causing very rare hang in SECC.

[rsimmonsjr]

This is a race condition found by running the tests in a loop in bash.

$ while cargo test; do true; done

Occasionally the channel just hangs. Most likely having to do with the Condvar signaling on waiting for available message or space to enqueue message. This issue is to debug and solve that.

[FenrirWolf]

I'm able to reproduce the hang pretty reliably with this monte carlo example in axiom.

When I run the example in LLDB I get this trace for the dispatcher thread and this trace for the send_after thread.

I also added some debug prints to Axiom. I took these lines here and changed them like so:

while !system.data.shutdown_triggered.load(Ordering::Relaxed) {
    if let Ok(actor) = receiver.receive_await_timeout(thread_timeout) {
        Actor::receive(actor);
    } else {
        println("Else branch reached");
    }
    println!("looping");
}

Which resulted in this output to stdout.

[rsimmonsjr]

Fixed.