feat(external): add WaveSurfer audio widget demo (#41)

## Summary

Adds an interactive audio waveform widget demo using wavesurfer.js and
anywidget.

Note: If the idea of the notebook is relevant - i'll be happy to work on
it - otherwise - you can reject it and close it. I could try to use
https://github.com/pengzhendong/wavesurfer instead of the directly the
js library.

## Features

- 🎵 **Interactive Waveform**: Draggable/resizable regions for audio
segment annotation
- ⏩ **Playback Speed**: 7 speed options (0.5x - 2.0x) for
transcription/analysis
- 🎹 **Synthetic Audio**: Self-contained with multi-pitch A major chord
sample
- 📁 **File Upload**: Optional upload support (MP3/WAV/OGG/M4A)
- 🔄 **Bidirectional Sync**: Python ↔ JavaScript state via traitlets
- 📊 **Table Integration**: Two-way sync between table and waveform
- 📥 **Export**: JSON export of annotated regions

## Use Cases

- Audio transcription and annotation
- Podcast chapter marking
- Music analysis (verse/chorus identification)
- Language learning with sentence-level audio alignment
- Torah reading verse boundary alignment

## Technical Details

- **Widget Framework**: anywidget (portable across Jupyter/marimo)
- **Audio Library**: wavesurfer.js v7 (loaded from CDN)
- **Dependencies**: PEP 723 compliant (marimo, anywidget, traitlets,
numpy)
- **Self-Contained**: Generates 5-second synthetic audio (A4/C5/E5
chord)
- **Size**: 18KB, 17 cells, ~600 lines
- **Python**: 3.12+ compatible

## Testing

- [x] \`uvx marimo check\` passes
- [x] Session metadata generated successfully
- [x] Notebook runs without errors
- [x] All interactive features tested

Author: sephib