TokenizingByCharacters export to Onnx

[kere-nel]

• Transformer that tokenizes by character and returns the characters (as uint16)
• Since there's not a comparable onnx operator, a label encoder is used to map a string token to it's corresponding character value. This will unfortunately make the model much larger, since 65535 values have to be saved as a mapping guide for label encoder.

[harishsk]

/// | Exportable to ONNX | No |

Please change this line whenever you add new support

---

Refers to: src/Microsoft.ML.Transforms/Text/TokenizingByCharacters.cs:610 in 0532330. [](commit_id = 0532330, deletion_comment = False)