chore(deps): cap torch ceiling at <2.12, pin Linux wheels to cu128 (#3570)

* chore(deps): ceiling + cuda

* ci: bump cuda version docker image

* ci: add cpu wheel to release workflow

* chore(deps): update uv.lock

* docs: update installation with cuda note

docker/Dockerfile.benchmark.robotwin

@@ -35,7 +35,7 @@ USER root
 ARG ROBOTWIN_SHA=0aeea2d669c0f8516f4d5785f0aa33ba812c14b4
 RUN apt-get update \
     && apt-get install -y --no-install-recommends \
-         cuda-nvcc-12-6 cuda-cudart-dev-12-6 \
+         cuda-nvcc-12-8 cuda-cudart-dev-12-8 \
          libvulkan1 vulkan-tools \
     && mkdir -p /usr/share/vulkan/icd.d \
     && echo '{"file_format_version":"1.0.0","ICD":{"library_path":"libGLX_nvidia.so.0","api_version":"1.3.0"}}' \

docker/Dockerfile.internal

@@ -18,7 +18,7 @@
 # docker build -f docker/Dockerfile.internal -t lerobot-internal .
 
 # Configure the base image for CI with GPU access
-ARG CUDA_VERSION=12.6.3
+ARG CUDA_VERSION=12.8.1
 ARG OS_VERSION=24.04
 FROM nvidia/cuda:${CUDA_VERSION}-base-ubuntu${OS_VERSION}