As an Intel GPU Researcher I would like elegant handling of ZE_AFFINITY_MASK

[lhlawson]

Is your feature request related to a problem? Please describe.
I'm always frustrated when the GEOPM service & runtime consume different ZE_AFFINITY_MASK values

Describe the solution you'd like
A helper object to resolve ZE_AFFINITY_MASK mapping conflicts between user & service values.

Describe alternatives you've considered
No change - not handling ZE_AFFINITY_MASK issues

Additional context
There is a potential issues where the service may cache number of GPUs based upon no affinity mask and receive N gpus, while the user sets ZE_AFFINITY_MASK=0 and expects 1 GPU. Or even sets 0.0 and expects a single tile of a GPU.