feat(aws): add speaker diarization support to STT

[huuthieu]

This commit adds speaker diarization capability to the AWS Transcribe STT plugin:

1. Enable diarization capability in STTCapabilities when show_speaker_label is set to True
2. Extract speaker_id from AWS Transcribe response items and populate it in SpeechData

When show_speaker_label=True is passed to the STT constructor, the plugin will now:

• Advertise diarization capability to the LiveKit framework
• Extract speaker labels (e.g., "spk_0", "spk_1") from AWS Transcribe responses
• Include speaker_id in the SpeechData returned to the application

Usage example:

stt = aws.STT(
    language="en-US",
    show_speaker_label=True,
)

async for event in stt_stream:
    if event.type == stt.SpeechEventType.FINAL_TRANSCRIPT:
        speaker_id = event.alternatives[0].speaker_id
        text = event.alternatives[0].text
        print(f"Speaker {speaker_id}: {text}")

[CLAassistant]

All committers have signed the CLA.