Stop using google_compute_image data source

[mr0re1]

Due to hashicorp/terraform-provider-google#21766 stop using data "google_compute_image".

• Following validations has been removed:
  • Check that boot disk size >= image size;
  • Check that image license is aligned with allow_automatic_updates.
• Move duplicate source_image_logic.tf into community/modules/internal/slurm-gcp/image_logic;
• Remove default instance_image values from user facing modules.

[rohitramu]

I think we should wait on this PR until we get a response on hashicorp/terraform-provider-google#21766, for a few reasons:

• The failure mode "disk size is less than image size" is difficult to identify and diagnose in a deployment
• Removing the default image value almost certainly makes the user experience worse - a beginner Cluster Toolkit user may pick a non-Slurm image and be confused as to why nothing's working
• Image deletion seems like a rare problem with an easy workaround (don't delete in-use images)

[rohitramu]

Discussed offline.

• I still feel that the failure mode "disk size is less than image size" is difficult to identify and diagnose in a deployment. We should add this check back when data "google_compute_image" fails on destroy path if image doesn't exist hashicorp/terraform-provider-google#21766 is resolved
• A default image will still be selected if one isn't provided - the specific image just isn't defined in the nodeset or login node settings. This shouldn't have as big an impact on user experience.