Fix max_new_tokens for assistant model in assistant generation

[jmamou]

What does this PR do?

During assistant generation, at each iteration, assistant model generate num_assistant_tokens tokens.

If the maximum numbers of tokens to generate is limited by max_len, in the case that max_len-cur_len is less than num_assistant_tokens, it will be more efficient for the assistant model to generate only max_len-cur_len instead of generating num_assistant_tokens.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

@gante
@echarlaix

[jmamou]

It seems that it has been solved yesterday by @ofirzaf #28508 (comment)