docs(examples): add microphone live transcription example

[GregHolmes]

Goal

Provide a maintained, copy-pasteable Python example that streams microphone audio to Listen V1 without adding microphone capture or platform audio dependencies to the core SDK.

Scope

• Add examples/18-transcription-live-microphone.py using the optional sounddevice dependency.
• Explain installation, device selection, and audio format settings in the script.
• Read DEEPGRAM_API_KEY from the environment and use DeepgramClient plus the public Listen V1 socket API.
• Send linear16 microphone bytes at the matching sample rate and channel count.
• Handle transcript, error, close, Ctrl-C, and CloseStream paths cleanly.
• Add the example to examples/README.md.

Non-goals

• No microphone code, optional extra, or audio-device dependency in deepgram-sdk.
• No cross-platform device abstraction beyond what sounddevice provides.
• No automatic reconnection or application-specific audio pipeline.

Acceptance

• The script imports only after the user installs sounddevice.
• It runs against a real microphone with documented setup.
• It uses public SDK APIs only and does not expose credentials.