fix: apply channel averaging correctly in audio feature extractors

[jonghwanhyeon]

What does this PR do?

This PR fixes incorrect channel averaging in the following audio feature extractors:

• CohereAsrFeatureExtractor
• ParakeetFeatureExtractor
• Phi4MultimodalFeatureExtractor
• LasrFeatureExtractor
• VoxtralRealtimeFeatureExtractor

The bug had two parts:

• batched tensor inputs shaped (batch, channels, length) were averaging the last dimension instead of the channel dimension.
• batched sequence inputs were just computing speech.mean(...) without writing the result back into raw_speech.

This change averages batched tensor inputs across channels with mean(1) and writes the mono-converted sample back to raw_speech for sequence inputs.

Code Agent Policy

The Transformers repo is currently being overwhelmed by a large number of PRs and issue comments written by
code agents. We are currently bottlenecked by our ability to review and respond to them. As a result,
we ask that new users do not submit pure code agent PRs at this time.
You may use code agents in drafting or to help you diagnose issues. We'd also ask autonomous "OpenClaw"-like agents
not to open any PRs or issues for the moment.

PRs that appear to be fully agent-written will probably be closed without review, and we may block users who do this
repeatedly or maliciously.

This is a rapidly-evolving situation that's causing significant shockwaves in the open-source community. As a result,
this policy is likely to be updated regularly in the near future. For more information, please read CONTRIBUTING.md.

• I confirm that this is not a pure code agent PR.

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?

@eustlb @ebezzam @vasqu

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: cohere_asr, lasr, parakeet, phi4_multimodal, voxtral_realtime

[eustlb]

Hey thanks a lot for your PR, there is indeed a bug here, but this bug is rather due to ambiguous representation of stereo audio across the lib.

Your fix is correct in the following assumption: stereo audio is in shape [BS, channels, audio lenght]. Were we to pass a [channels, audio lenght], then there is confusion between batch size and channel dim.

Anyway, this is unclear and should be covered by #44394, so I am going to alt review here just a bit to think about the best way to solve these ambiguities and come back to you asap.