Tensura: Voice Cast 0.2.1

Fixed voicecasting failing on fresh installations with IllegalStateException: English model missing.

• Bundled the English speech-recognition model directly inside the mod JAR.
• Added automatic local model setup when voice listening first starts. No separate model download or manual installation is needed with the default settings.
• Model setup runs in the background, and speech recognition remains fully offline.
• Preserved existing models, aliases, microphone settings and listening preferences.
• Added clearer instructions for missing custom models and incomplete model folders.

Updating: Close Minecraft and replace the old Voicecasting JAR with version 0.2.1. Keep only one version installed. Equip a TenSura ability, run /tvoice listen or /tvoice auto on, close chat and wait for Listening locally before speaking.

Note: The JAR is larger because it now includes the English recognition model.