Add voice `model` / `model.id` / `model.parameters` configuration to the voice recipe

Summary

The modality voice block supports selecting and tuning the synthesized voice model via an outbound.model section, but our recipes don't demonstrate it. We should show how to set model, model.id, and/or model.parameters so developers can pick the right voice model and tune delivery (speed, stability, similarity).

Based on the docs: Configure Voice in Agent Script

Suggested Solution

Add an example under the modality voice block's outbound section:

modality voice:
    outbound:
        persona_id: "hash_id_here"
        model:
            id: "eleven_flash_v2_5"
            parameters:
                speed: 1.0
                stability: 0.8
                similarity: 0.75

Cover the supported model IDs and when to use each:

And the model.parameters (support varies by model):

• speed (0.7–1.2) — how fast the voice speaks
• stability (0–1) — how consistent the delivery is
• similarity (0–1) — how closely output matches the original voice's character

Notes

Related to the broader voice recipe in #95 and the per-subagent model configuration in #94. This issue is scoped specifically to the voice outbound.model configuration.

Code of Conduct

• I agree to follow this project's Code of Conduct