Bug 1788904

Description of problem:

Our official docs available here https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/14/html/instances_and_images_guide/ch-virtual-gpu#nvidia-build-guest-image engages utils like genisoimage and virt-customize to achieve something what OpenStack has been made for. 

Instead of customising qcow2 image file in-place on Director I think it would be better to re-write the procedure to use OpenStack's capabilities to do it. This would be easier approach and will work better for cases where Director's node is virtualised (avoid nested virtualisation). 

The procedure would be something like the following:
1) Upload stock RHEL image to glance
2) Spin-up an instance from it
3) Register it with RHN
4) Update system and install extra packages
4) Upload NVIDIA drivers there, unarchive and install
5) Prepare gridd.conf file
6) Shutdown the instance
7) Create image from the instance

IMHO this is less hacky approach comparing to preparation of ISO file and shell script to modify qcow2 image. 

Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info: