Use the nvidia-supplied nvidia-ml-py instead of pynvml

[mattip]

core/requirements.txt

Line 16 in e1eaf26

pynvml>=11.0.0,<11.5

This project depends on pynvml, where the nvidia-blessed bindings to nvml are in the nvidia-ml-py package instead. This is causing problems for the gpustat project, which needs some internals from nvidia-ml-py to properly support older nvml version. Would it be possible to switch?