feat: Add Whisper transcription support with speech_recognition fallback

[AlexZhangji]

Changes

• Added OpenAI Whisper support for audio transcription in WavConverter and Mp3Converter
• Implemented automatic fallback to speech_recognition if Whisper fails or non-OpenAI client passed in.
• Added audio transcription tests (commented out by default) with a short test.wav.

Usage

from markitdown import MarkItDown
from openai import OpenAI

client = OpenAI()  
md = MarkItDown(llm_client=client)

result = md.convert("audio.wav")  # or "audio.mp3"
print(result.text_content)
# Output: ### Audio Transcript (Whisper):
#         <transcription content in markdown>

[AlexZhangji]

@microsoft-github-policy-service agree