Add voice input (speech-to-text) to HumanCLI for Go2

Issue

Add voice input capability to HumanCLI so users can speak commands instead of typing them. Validate that it works on the Unitree Go2 platform.

Requirements

• Implement speech-to-text using Python audio libraries (e.g., pyaudio, speech_recognition, or similar)
• Add directly to HumanCLI module (dimos/agents/cli/human.py)
• Support microphone input on Go2
• Handle audio device selection/configuration
• Test on actual Go2 hardware

Implementation Considerations

• Use lightweight STT that runs on-device or can call external API
• Handle noise/background audio on robot
• Provide fallback to typed input if voice fails
• Toggle for enabling/disabling voice input

Acceptance Criteria

• Voice input works on Go2
• User can speak commands to the agent
• Transcription is accurate enough for navigation/control commands
• Graceful fallback to text input if voice unavailable

Related

• Pair with: #1273 (Voice output TTS)

Start in dimos/agents/cli/human.py and review the HumanCLI input flow before choosing the Python audio or speech-to-text approach. Validate microphone selection, voice-to-text commands, typed fallback, and the enable/disable toggle on actual Unitree Go2 hardware; completion requires the listed voice-input acceptance criteria to work there.