feat: Add Whisper STT Extension using faster-whisper

[Nsuccess]

Description

This PR adds a Whisper STT (Speech-to-Text) extension for the TEN Framework using the faster-whisper library.

Closes #1969

Features

• ✅ Complete ASR Implementation: Inherits from AsyncASRBaseExtension following TEN Framework patterns
• ✅ Optimized Performance: Uses faster-whisper (4x faster than openai/whisper)
• ✅ Multiple Model Sizes: Support for tiny, base, small, medium, large-v1/v2/v3
• ✅ CPU & GPU Support: Configurable device and compute types (int8, float16, float32)
• ✅ Multi-Language: 99+ languages with automatic detection
• ✅ Translation: Translate speech to English
• ✅ VAD Filtering: Built-in voice activity detection using Silero VAD
• ✅ Production-Ready: Auto-reconnection, audio dumping, standardized logging
• ✅ Buffer Strategy: Keep mode with 10MB limit for timestamp accuracy

Implementation Details

Architecture

• Extension: WhisperSTTExtension - Main ASR extension class
• Client: WhisperClient - Handles faster-whisper model and inference
• Config: WhisperSTTConfig - Pass-through params design for flexibility
• Reconnection: Exponential backoff with configurable max attempts

Files Added (14 files, 1,324 lines)

• whisper_stt_python/extension.py - Main extension implementation
• whisper_stt_python/whisper_client.py - Faster-whisper client wrapper
• whisper_stt_python/config.py - Configuration management
• whisper_stt_python/reconnect_manager.py - Auto-reconnection logic
• whisper_stt_python/addon.py - Extension entry point
• whisper_stt_python/const.py - Constants
• whisper_stt_python/manifest.json - Extension metadata
• whisper_stt_python/property.json - Default configuration
• whisper_stt_python/requirements.txt - Dependencies
• whisper_stt_python/README.md - Comprehensive documentation
• whisper_stt_python/tests/test_config.py - Config tests (10 tests)
• whisper_stt_python/tests/test_extension.py - Extension tests (15 tests)

Testing

• ✅ 25 Unit Tests: Full coverage with mock-based testing
• ✅ Config Tests: Default values, JSON parsing, sensitive masking, language normalization
• ✅ Extension Tests: Initialization, connection, audio sending, finalize, callbacks
• ✅ No Real API Calls: All tests use mocks for fast, reliable execution

Configuration Example

{
  "params": {
    "model": "base",
    "device": "cpu",
    "compute_type": "int8",
    "language": "en",
    "task": "transcribe",
    "sample_rate": 16000
  }
}