feat(cli): speech-to-text v2 — streaming

[github-actions]

Adds the streaming recipe for speech-to-text v2 (cli).

Closes part of #489

[github-actions]

❌ CLI Recipe Tests

── recipes/cli/speech-to-text/v2/streaming ──
FAIL: example.sh exited with status 2
/home/runner/work/recipes/recipes/recipes/cli/speech-to-text/v2/streaming/example.sh: line 14: ffmpeg: command not found
✓ Streaming stdin · flux-general-en · send EOF (Ctrl+D) to finish
ERROR: Command failed: server rejected WebSocket connection: HTTP 400
Error: server rejected WebSocket connection: HTTP 400
✗ FAILED

[claude]

@deepgram/devrel — VP escalation: PR #490 has had failing E2E checks (test-cli FAILURE) since 2026-04-05 (~2 days) with no fix attempt. The lead-fix workflow does not exist and lead-test dispatch returned 403. Manual review needed.

State: labels=[type:samples, language:cli], test-cli=FAILURE, e2e-api-check=FAILURE, no reviews.

[claude]

@deepgram/devrel — VP escalation: PR #490 has been stuck for >4 hours and re-trigger attempts have not resolved it. Manual review needed.

State: labels=[type:samples, language:cli], test-cli and e2e-api-check FAILED on 2026-04-05 with no fix activity since. Workflow dispatch returned 403.