An Investigation of Streaming Non-Autoregressive sequence-to-sequence Voice Conversion