fix(net): keep track::Fetching unwind safe

rs/moq-net/src/model/group.rs

@@ -1129,7 +1129,10 @@ pub struct Consumer {
 }
 
 /// Subscriber-specific policy for expiring a group after it was handed out.
-pub(crate) trait Expiry: Send + Sync {
+///
+/// Unwind safe so the [`Consumer`] holding it is, and so the published
+/// [`track::Fetching`] that holds a consumer stays unwind safe too.
+pub(crate) trait Expiry: Send + Sync + std::panic::UnwindSafe + std::panic::RefUnwindSafe {
 	/// Return whether the group is stale, registering `waiter` for anything that
 	/// could change the answer while the group remains live.
 	fn is_expired(&self, waiter: &kio::Waiter) -> bool;

rs/moq-net/src/model/track.rs

@@ -8080,6 +8080,15 @@ mod test {
 		assert_eq!(retry.await.unwrap().sequence, 5);
 	}
 
+	/// Dropping an auto trait from a published type is a semver break, so the group
+	/// consumer a cached fetch holds must not cost `Fetching` its unwind safety.
+	#[test]
+	fn fetching_is_unwind_safe() {
+		fn assert_unwind_safe<T: std::panic::UnwindSafe + std::panic::RefUnwindSafe>() {}
+		assert_unwind_safe::<Fetching>();
+		assert_unwind_safe::<group::Consumer>();
+	}
+
 	/// A fetch that hits the cache holds the group until polled, so a handler that aborts
 	/// the group once nobody wants it (an abandoned upstream fetch) leaves it alone.
 	#[tokio::test]