[design] Speech to text and text to speech nodes in pipelines

[snopoke]

Related to #1137

Write up a proposal for how to handle TTS and STT in pipelines.

In legacy bots it is handled in the Channels (https://github.ctom/dimagi/open-chat-studio/blob/08b37e392b229a4d6ff432347da403679881def0/apps/chat/channels.py#L430 &

open-chat-studio/apps/chat/channels.py

Line 396 in 08b37e3

def _reply_voice_message(self, text: str):

).

This approach still works with pipelines but the pipeline should be responsible for handling this to make it more flexible - maybe you want to use Whisper or maybe you are using a multi-modal modal that will accept the attachment directly