dsh-wsl-gpu: GPU Visibility and Inference Port Detection Tool in WSL

The `dsh-wsl-gpu` plugin for DeepSeek Harness is designed to address VRAM conflicts and resource bottlenecks in local large model inference on WSL2. By parsing `nvidia-smi` output, it offers GPU status monitoring, Blackwell architecture hints, and inference port scanning to help developers quickly diagnose resource utilization. It can be installed via Kit or enabled through the DSH manager, requires DSH version >= 0.1.2, and is suitable for pre-driver-update, pre-build-failure, and pre-model-loading resource checks.