PhysicalBrdige: Always perform "last read" check in heartbeat

When we have slowly adding things to the heartbeat (originally intended just to send data to keep connections alive) like detecting connection health, the if/else has gotten more complicated. With the addition of `HeartbeatConsistencyChecks`, we prevented some fall throughs to later checks which means that if that option is enabled, we were no longer detecting dead sockets as intended.

This is a tactical fix for the combination, but I think overall we should look at refactoring how this entire method works because shoehorning these things into the original structure/purpose has been problematic several times.

src/StackExchange.Redis/PhysicalBridge.cs

@@ -652,7 +652,9 @@ internal void OnHeartbeat(bool ifConnectedOnly)
  // so if we have an empty unsent queue and a non-empty sent queue, test the socket.
  KeepAlive();
  }
- else if (timedOutThisHeartbeat > 0
+
+ // This is an "always" check - we always want to evaluate a dead connection from a non-responsive sever regardless of the need to heartbeat above
+ if (timedOutThisHeartbeat > 0
  && tmp.LastReadSecondsAgo * 1_000 > (tmp.BridgeCouldBeNull?.Multiplexer.AsyncTimeoutMilliseconds * 4))
  {
  // If we've received *NOTHING* on the pipe in 4 timeouts worth of time and we're timing out commands, issue a connection failure so that we reconnect