Fix audio-related config naming for Gemma3n

[ywang96]

What does this PR do?

While working on adding multimodal support for Gemma3n on vLLM, I noticed audio_soft_tokens_per_image should really be renamed to audio_soft_tokens_per_audio. This PR updates the variable naming but we'll also need to update this field in the model repository.

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?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[ywang96]

I've made corresponding model config PRs in all model repositories

https://huggingface.co/google/gemma-3n-E4B-it/discussions/12
https://huggingface.co/google/gemma-3n-E4B/discussions/3
https://huggingface.co/google/gemma-3n-E2B-it/discussions/8
https://huggingface.co/google/gemma-3n-E2B/discussions/3

Maybe @zucchini-nlp you can take a look and let me know what you think? Thanks!

[zucchini-nlp]

Yeah, makes total sense to me!

Though as the model was released last week in v4.53.0 and probably we have already several tuned copied of it on the hub, this should be done in a non-breaking way with a deprecation cycle (imo 1 release is enough)