Add decord/librosa to dependency (audio processing)

[hengtaoguo]

Description

Following guides, this PR adds decord, librosa to requirements.txt, and regenerated tpu/cuda12 dependencies using JAX hash 189845a86062e22afbed880d4f4d31a38a4e5786 (link, 01/16/2026).

Note: This PR automatically spins up:

• JAX: version from 0.8.1 to 0.8.3
• libtpu: from 0.0.30 to 0.0.32

seed-env \
  --local-requirements=dependencies/requirements/base_requirements/tpu-base-requirements.txt \
  --host-name=MaxText \
  --seed-commit=189845a86062e22afbed880d4f4d31a38a4e5786 \
  --python-version=3.12 \
  --requirements-txt=tpu-requirements.txt \
  --output-dir=generated_tpu_artifacts

seed-env \
  --local-requirements=dependencies/requirements/base_requirements/gpu-base-requirements.txt \
  --host-name=MaxText \
  --seed-commit=189845a86062e22afbed880d4f4d31a38a4e5786 \
  --python-version=3.12 \
  --requirements-txt=cuda12-requirements.txt \
  --hardware=cuda12 \
  --output-dir=generated_gpu_artifacts

Tests

Locally created a new venv and dependency installation was successful:

uv venv --python 3.12 --seed maxtext_venv
source maxtext_venv/bin/activate

cd maxtext
uv pip install -e .[tpu] --resolution=lowest
install_maxtext_github_deps

CI tests.

Checklist

Before submitting this PR, please make sure (put X in square brackets):

• I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.