sherpa-onnx-offline-tts: Allow piping audio from tts directly to stdout for further processing

[domasofan]

Hi all,

It would be nice if it would be possible to add this feature so audio can be piped directly for example into ffmpeg or lame.
So generating a wav file first can be skipped.
Most of the tts audio i create need to be converted into a compressed format and so i can skip this step and safe a little bit of my ssd live. :-)

Greetings and thanks,
Simon