fix: include torchaudio for local ASR and document requirement

[MirxaWaqarBaig]

Add torchaudio==2.11.0 to runtime dependencies.
Document torchaudio requirement for storyline.local_asr in English and Chinese README install sections.
Fixes runtime failure: ModuleNotFoundError: No module named 'torchaudio' during local ASR execution.

[MirxaWaqarBaig]

Validation steps

Environment

• OS: Ubuntu on WSL2
• Python: 3.11 (conda env: storyline)

Repro before this fix

1. Run local ASR flow (storyline.local_asr).
2. It failed with:
   ModuleNotFoundError: No module named 'torchaudio'
   (from funasr import path in asr_node).

Validation after this fix

1. Install from updated requirements:
   pip install -r requirements.txt
2. Confirm:
   python -c "import torchaudio; print(torchaudio.version)"
3. Re-run local ASR workflow in app.
4. local_asr executes successfully and downstream speech_rough_cut can proceed.

Docs

• README.md and README_zh.md now explicitly mention torchaudio requirement for storyline.local_asr.