[Trivial] Extend driver shutdown time (#2726)

# Description
We have been seeing panics like
[this](https://cowservices.slack.com/archives/C037PB929ME/p1715586375609859)
sporadically on system restarts.

I believe the root-cause for this is that the time we leave the driver
to gracefully shut down (10s) is shorter than the time a single solve
request may be taking. This when a solve request is in flight, we don't
wait long enough before forcing shutdown. Timeline from the example ☝️

- 2024-05-13T07:45:28.781Z driver competition started
- 2024-05-13T07:45:30.313Z likely receipt of shutdown signal (no logs)
- 2024-05-13T07:45:40.313Z panic due to timeout
- 2024-05-13T07:45:43.254Z driver competition deadline would have been
reached

# Changes
- [x] Wait 20s for timeout
- [x] Log when shutdown signal is received

## How to test
CI and hope to no longer see those panics.

crates/driver/src/run.rs

@@ -79,8 +79,10 @@ async fn run_with(args: cli::Args, addr_sender: Option<oneshot::Sender<SocketAdd
     tokio::select! {
         result = &mut serve => panic!("serve task exited: {result:?}"),
         _ = shutdown_signal() => {
+            tracing::info!("Gracefully shutting down API");
             shutdown_sender.send(()).expect("failed to send shutdown signal");
-            match tokio::time::timeout(Duration::from_secs(10), serve).await {
+            // Shutdown timeout needs to be larger than the auction deadline
+            match tokio::time::timeout(Duration::from_secs(20), serve).await {
                 Ok(inner) => inner.expect("API failed during shutdown"),
                 Err(_) => panic!("API shutdown exceeded timeout"),
             }