Prevent double-trigger: guard do_hide/do_recover + drain stale audio

Redesign launcher landing page with cartoon comic strip and full guide

Fix short_description length for HF validation

Add full README with usage guide, voice commands, and install instructions

Silence mic output while hiding, add voice recovery commands

Use goto_sleep()/wake_up() for proper sleep pose with built-in sounds and animation

Grammar-constrained recognizer for speed, add boo/peek/hide/go as triggers

Show voice_status in UI — exact step visible in browser regardless of logging

Fix stereo->mono conversion (mean axis=1 not flatten), add start_playing() per SDK docs

Fix logger namespace to reachy_mini.*, wrap voice_loop to surface crashes at ERROR level

Bump diagnostics to WARNING level, show voice-ready state in UI

Switch all print() to logger.info() so diagnostics appear in launcher logs

Add verbose diagnostics, simulation detection, and sim UI notice

Switch to vosk (offline, no FLAC), streaming per-chunk recognition

Fix audio API: use reachy_mini.media.get_audio_sample() not .media.audio, add pipeline warmup sleep

Use robot mic via reachy_mini.media.audio (gstreamer_no_video), drop sounddevice

Move speech recognition to Python (sounddevice+SpeechRecognition), simplify browser UI

Implement boo-triggered hide: voice detection, 30s auto-recover, dark UI

Initial commit