Fix race-condition-induced dead lock on replay restart in 1.14+

MC clears it scheduled tasks on disconnect. If we don't wait for
that to have happened on the async sender, then we might loose
some of the initial handshake packets (resulting in a dead lock in
fabric's registry syncing code).

src/main/java/com/replaymod/replay/ReplayHandler.java

@@ -35,6 +35,8 @@ import net.minecraft.network.ClientConnection;
 
 import java.io.IOException;
 import java.util.*;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeoutException;
 
 //#if MC>=11300
 import com.replaymod.replay.mixin.EntityLivingBaseAccessor;
@@ -162,18 +164,22 @@ public class ReplayHandler {
         channel.close();
 
         // Force re-creation of camera entity by unloading the previous world
-        ReplayMod.instance.runSync(() -> {
-            //#if MC>=11300
-            mc.mouse.unlockCursor();
-            //#else
-            //$$ mc.setIngameNotInFocus();
-            //#endif
-            //#if MC>=11400
-            mc.disconnect();
-            //#else
-            //$$ mc.loadWorld(null);
-            //#endif
-        });
+        try {
+            ReplayMod.instance.runSync(() -> {
+                //#if MC>=11300
+                mc.mouse.unlockCursor();
+                //#else
+                //$$ mc.setIngameNotInFocus();
+                //#endif
+                //#if MC>=11400
+                mc.disconnect();
+                //#else
+                //$$ mc.loadWorld(null);
+                //#endif
+            });
+        } catch (InterruptedException | ExecutionException | TimeoutException e) {
+            LOGGER.error("Failed to properly restart (shutdown) replay:", e);
+        }
 
         restrictions = new Restrictions();