Add method to get request ID from TTS requests

[bd-g]

Proposed changes

The Deepgram request ID for TTS requests is passed via a header value, not in the output data (which is the audio waveform for the speech). We should expose a top level function to be able to access the request ID, with the SDK handling pulling this from the headers.

Context

Recorded request IDs is important for debugging and general tracking purposes.