Add Vulkan Docker image support

[0xShug0]

Resolves #459

Summary

• add a Vulkan Dockerfile for full audio.cpp builds
• publish full-vulkan Docker images in the existing Docker workflow
• document Vulkan Docker build/run commands, including /dev/dri and render/video group access

Validation

Built the local Vulkan image and verified the runtime image contains the CLI, server, model manager, model_perf, and libggml-vulkan.so. Also verified libggml-vulkan.so links against libvulkan.so.1.

End-to-end checks inside the Vulkan Docker image:

Case	Result
Soprano TTS CLI	generated WAV successfully; logs show Vulkan backend/device
Qwen3 ASR CLI	transcribed the sample correctly; logs show Vulkan backend/device
PocketTTS CLI	generated WAV successfully; logs show Vulkan backend/device
Server HTTP TTS	/health reported backend=vulkan and /v1/audio/speech returned WAV

One deployment detail from validation: the default container user needs the host render/video groups passed through, otherwise Vulkan device enumeration may see no devices.