Optimizing LoRa for Edge Computing with TinyML Pipeline for Channel Hopping