feat(python): voice-agents v1 — proactive-speech

[github-actions]

Adds the proactive-speech recipe for voice-agents v1 (python).

Demonstrates injecting agent messages during user silence using
send_inject_agent_message with behavior="queue" so the agent
fills conversational pauses with context-aware prompts — gentle
patience, offer clarification, and suggest alternatives — without
interrupting if the user begins speaking.

Closes #571

[claude]

@deepgram/devrel — VP escalation: PR #574 has been stuck for >7 days with no CI checks triggered. Workflow re-trigger not possible due to GITHUB_TOKEN permissions. Manual review needed.

State: labels=[type:samples, language:python], checks=none

[claude]

@deepgram/devrel — VP escalation: PR #574 has been stuck for >3 days with no CI checks running. Workflow dispatch triggers failed (HTTP 403 — GITHUB_TOKEN limitation). Manual intervention needed to trigger tests or merge.

State: no checks, no reviews, last updated 2026-06-09

[claude]

@deepgram/devrel — VP escalation: PR #574 has been stuck for >59 hours. Workflow re-trigger failed (GITHUB_TOKEN 403). Manual review needed.

State: Labels: type:samples, language:python. Checks: NONE

[claude]

@deepgram/devrel — VP escalation: PR #574 has no CI checks and has been stale for >7 days. Workflow dispatch is blocked by token permissions. Manual re-trigger of CI or review needed.

State: labels=[language:python, type:samples], checks=none