Text this: The Subword‐Character Multi‐Scale Transformer With Learnable Positional Encoding for Machine Translation