[FEATURE] - Support open-ai Realtime API

[niztal]

Description:

Open AI just announced the new GPT 4o Realtime API, which brings superb capabilities of Speech to Speech, Speech to Text, Text to Speech combining functions/tools as well.

Seems like this model is based on web socket openai's API.

I really think combining this capabilities would gain huge value to flowise having the option to build RAG+commands voice applications

Additional Info:

1. Announcement
2. API doc

@HenryHengZJ

Thanks