Add voice-command router (injectable speech-to-text)

[JE-Chen]

Hands-free batch (approved under 批准全部). Full layers + tests + EN/Zh v44 docs + README.

Feature (utils/voice, pure-stdlib)

• VoiceRouter: maps spoken trigger phrases to AC_* action lists; feed it recognized text and it runs the closest registered command. Phrase matching reuses the project's fuzzy matcher (Add fuzzy string matching and dedupe #248), so "save the file" fires a "save file" command despite recogniser noise.
• Speech-to-text is out of scope and injectable: the router takes already-recognised text; a real mic/Vosk recogniser is a recognizer callable to listen_once, and the action runner is injectable too — so routing is fully unit-tested with no audio or speech dependency. The env-bound STT path stays outside the framework, honestly.
• Executor AC_voice_register / AC_voice_dispatch / AC_voice_list / AC_voice_clear (actions as list or builder JSON string); MCP ac_voice_*; Builder under Agent.

Verification

• 10 tests pass (register/replace/phrases, fuzzy match + below-threshold None, dispatch runs matched actions via injected runner, no-match runs nothing, listen_once uses recognizer, executor round-trip, wiring, facade); ruff clean; radon no CC≥C; bandit clean; PySide6-free.
• Stacked on locale (Add locale-aware number/currency/date parsing #251); rebased cleanly onto dev.

[codacy-production]

Up to standards ✅

🟢 Issues 0 issues

Results:
0 new issues

View in Codacy

🟢 Metrics 41 complexity · 0 duplication

Metric	Results
Complexity	41
Duplication	0

View in Codacy

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud