Investigating the Reordering Capability in CTC-based Non-Autoregressive End-to-End Speech Translation