Voice2Text AI 0.4 native Linux rewrite

[crhy]

Summary

• replace the monolithic Qt/PyAudio application with GTK 4 and libadwaita
• use GStreamer for responsive microphone capture and streamed speech playback
• support system, light, and dark appearance modes
• stop dictation before Ask AI, Copy, and Clear actions
• preserve complete microphone names in Preferences
• stream Ollama responses and natural Edge TTS audio with offline eSpeak NG fallback
• eliminate stale X11/Compiz rendering artifacts
• rebuild the Flatpak on the GNOME 50 runtime
• add unit, lint, metadata, screenshot, and release validation

Validation

• Python compilation
• Ruff
• Pytest
• desktop entry validation
• Flatpak manifest and AppStream lint
• local Flatpak build, install, and repository lint
• manual dictation, microphone selection, Ollama, theme, clipboard, clear, and speech testing