Add support for nvidia-smi
Use the nvidia runtime in Telegraf containers to get access to GPU metrics.
The current node should be scanned for the presence of GPUs. If so, this configuration should be applied to Telegraf containers.
Edited by SIMONIN Matthieu