Documentation: in `step_tokenize()`, `udpipe` engine also takes `training_options`

[petrbouchal]

The problem

This is purely about documentation.

The documentation for step_tokenize(), under the parameter training_options says

A list of options passed to the tokenizer when it is being trained. Only applicable for engine == "tokenizers.bpe".

It also applies to udpipe as per https://www.emilhvitfeldt.com/post/textrecipes-version-0-4-0/.

It would be great if this could be added, otherwise operating udpipe via step_tokenize() is a bit of a mystery.

[EmilHvitfeldt]

Hello @petrbouchal 👋

This is very useful feedback! Please feel free to add other issues if other things are unclear!