Live voice assistance with prose and code

### What is the problem

So, deck is a presentation tool for agents and humans, and the thing that I feel that we can do, is to add voice to it, the basic version is definitely text to speech, but what I'm thinking is an upgraded version, we can use any agent toolkit and create live voice. The thing is that, when you're in office, you cannot really whisper to the agent, that's the reason I'm thinking of a two way communication. You can provide comments and reactions to the agent, when it speaks, those will be recorded as comments and if you're discussing on that, `deck wait` is going to provide the entire summary of what was exactly discussed between the agent and the human about it.

Imagine it like teams or slack huddle, someone is speaking and explaining you, while you can react and add comment, the agent answers that comment or note that to forward that.

Another thing is highlight, while explaining, I think this is another feature request, the agent can also highlight the piece of code, that we are looking at, or it want us to look at, and span of lines as tool when speaking, such that the explanation is much more intuitive.

### What you had in mind

_No response_

The issue names the `deck wait` command and describes live two-way voice, discussion summaries, reactions, and code highlighting, but it does not identify files, tests, or existing entry points. Start by mapping the CLI command and presentation flow, then determine the scope and acceptance criteria for voice, comments, summaries, and line highlighting before implementation.

Written by the indexing model from the issue text.