Correct OpenStack network configuration

We have been having weird network configurations with openstack-cheaha-internal and forced to update the routing table when starting a new instance.

https://gitlab.rc.uab.edu/rc/packer-openstack-hpc-image/-/blob/main/.gitlab-ci.yml?ref_type=heads#L271

added Sprint 24-3 label

assigned to @cmcclung

mentioned in merge request !77 (merged)

mentioned in commit 21117b85

closed with merge request !77 (merged)

We refactored the network configuration for this pipeline. The original configuration had some quirks that came about due to us still being in a "kicking the tires" phase of the environment when it was originally set up.

Access to UAB campus and Cheaha networks are required for this pipeline. The original network config had the cheaha-cluster network's router as the default route, which would provide access to the Cheaha network, but also prevent necessary traffic from routing to the internet. The actual req for this environment is for the uab-campus network's router to be the default route with the Cheaha router as a secondary static route.

During deployment, the OS's default route was being replaced with the uab-campus's gateway. We were troubleshooting intermittent network problems at instance deployment. This setup is not ideal, and could likely be a cause of the issues we were experiencing if the default route replacement via cloud-init userdata failed for any reason. In this case the instance would be inaccessible via network as we experienced.

To resolve this technical debt and rule out the routing change as a cause of the pipeline failures, we created a new network and modified the pipeline to use it. The new network has the uab-campus network's router as the default gateway. The router has a static route to the cheaha-cluster network's router. We're replacing the default dns config with Cheaha dns servers via userdata so the instance is able to mount GPFS volumes during boot.