NVIDIA CUDA runtime notice for optional Local AI GPU installation

Component: CUDA 12.4 runtime and cuBLAS DLLs from the pinned llama.cpp b9888
cudart archive. These proprietary files are downloaded only when the user
chooses a compatible NVIDIA GPU installation; they are not bundled in the
ViralText installer or portable package.

Official licence terms:
https://docs.nvidia.com/cuda/eula/index.html

Pinned archive:
https://github.com/ggml-org/llama.cpp/releases/download/b9888/cudart-llama-bin-win-cuda-12.4-x64.zip
SHA-256: 8c79a9b226de4b3cacfd1f83d24f962d0773be79f1e7b75c6af4ded7e32ae1d6

The NVIDIA CUDA Toolkit EULA and CUDA Supplement identify CUDA Runtime and
CUDA BLAS library files as distributable with qualifying applications, subject
to the agreement's conditions. This engineering notice is not a substitute
for the official terms or professional review of ToolTaktik's distribution.
The installer shows the official terms URL before download, and this notice is
retained beside the installed optional runtime.
