also route chat messages through the study batcher

since chat messages are versioned,
having them bypass the batcher results
in broken ordering, if other versioned
messages are being batched while the chat message goes through

src/main/scala/LilaHandler.scala

@@ -125,6 +125,8 @@ final class LilaHandler(
 
     _ match
       case t: TellRoomVersion => batcher.add(t.roomId, ClientIn.Versioned(t.json, t.version, t.troll))
+      case TellRoomChat(roomId, version, troll, payload) =>
+        batcher.add(roomId, ClientIn.Versioned(payload, version, troll))
       case LilaOut.RoomIsPresent(reqId, roomId, userId) =>
         lila.emit.study(LilaIn.ReqResponse(reqId, roomCrowd.isPresent(roomId, userId).toString))
       case LilaBoot => roomBoot(_.idFilter.study, lila.emit.study)