Add pre-start hook to all containers in container runtime to support GPU access

[Fizzbb]

check out the implementation of nvidia-container-runtime, modify runc/quark to support GPU access within container.

the hook will configure GPU access for the container by leveraging nvidia-container-cli from project libnvidia-container.

# Setup a new set of namespaces
cd $(mktemp -d) && mkdir rootfs
sudo unshare --mount --pid --fork

# Setup a rootfs based on Ubuntu 16.04 inside the new namespaces
curl http://cdimage.ubuntu.com/ubuntu-base/releases/16.04/release/ubuntu-base-16.04.6-base-amd64.tar.gz | tar -C rootfs -xz
useradd -R $(realpath rootfs) -U -u 1000 -s /bin/bash nvidia
mount --bind rootfs rootfs
mount --make-private rootfs
cd rootfs

# Mount standard filesystems
mount -t proc none proc
mount -t sysfs none sys
mount -t tmpfs none tmp
mount -t tmpfs none run

# Isolate the first GPU device along with basic utilities
nvidia-container-cli --load-kmods configure --ldconfig=@/sbin/ldconfig.real --no-cgroups --utility --device 0 $(pwd)

# Change into the new rootfs
pivot_root . mnt
umount -l mnt
exec chroot --userspec 1000:1000 . env -i bash

# Run nvidia-smi from within the container
nvidia-smi -L

https://developer.nvidia.com/blog/nvidia-docker-gpu-server-application-deployment-made-easy/
Dependency: nvidia-docker2 -> nvidia-container-toolkit -> libnvidia-container-tools -> libnvidia-container1