Automate the GPU Operator installation NVIDIA AI Enterprise special cases

[supertetelman]

NVIDIA AI Enterprise relies on special NVIDIA drivers that enable GPUs in a virtualized environment (vGPU). This is part of the product offerings that are delivered through NGC.

In order for NVIDIA AI Enterprise to function properly with the GPU Operator, there are a few special steps required. This includes using special helm packages for the operator, providing NGC keys to access private container resources, and using special configuration variables.

I propose we do the following:

• Break the current GPU Operator install path into a standard install and NVIDIA AI Enterprise install path
• Implement the NVIDIA AI Enterprise role tasks
• Introduce several new variables to support installing the NVIDIA AI Enterprise path and optionally, support the new NVIDIA License Server used in AI Enterprise (NLS)
• Document the new variables, install path, etc.

[ajdecon]

This should be addressed by #1059

[supertetelman]

We should tested this end-to-end today in master and it looks good!