Ignore key discard timer when detecting activity in unit tests

This caused rare flakes in at least
single_ack_eliciting_packet_triggers_ack_after_delay by classifying a
the connection as active for longer than expected.

Author: Ralith

quinn-proto/src/connection/mod.rs

@@ -3534,12 +3534,12 @@ impl Connection {
             .saturating_sub(self.path.in_flight.bytes)
     }
 
-    /// Whether no timers but keepalive, idle, rtt and pushnewcid are running
+    /// Whether no timers but keepalive, idle, rtt, pushnewcid, and key discard are running
     #[cfg(test)]
     pub(crate) fn is_idle(&self) -> bool {
         Timer::VALUES
             .iter()
-            .filter(|&&t| t != Timer::KeepAlive && t != Timer::PushNewCid)
+            .filter(|&&t| !matches!(t, Timer::KeepAlive | Timer::PushNewCid | Timer::KeyDiscard))
             .filter_map(|&t| Some((t, self.timers.get(t)?)))
             .min_by_key(|&(_, time)| time)
             .map_or(true, |(timer, _)| timer == Timer::Idle)