The auto_delete parameter (and therefore auto_delete_disk variable) within the NFS instance's boot_disk block is only effective when the instance implicitly creates the boot disk (i.e., disk properties are defined directly within the instance block using initialize_params). Even with auto_delete=false, the separate google_compute_disk resource goes through it's own destroy lifecycle and the disk is deleted - so ultimately, the variable currently does nothing.

This PR addresses reproducible issue #3566, which was introduced as a regression in bb47067

This PR addresses the auto_delete issue, and more broadly expands the data retention options by:

1. Removing the non-functioning auto-delete parameterization
2. Implementing an alternative data protection system, for both the boot disk and the data disk, via create_snapshot_before_destroy and create_boot_snapshot_before_destroy

Design notes:

1. Retained the existing, heavily-customized boot disk system (i.e., the null_resource, custom lifecycle management, and dedicated boot compute_disk specification), which blocks the ability to use the boot disk auto_delete control.
2. While there's no working functionality to deprecate, the auto_delete_disk variable is nevertheless deprecated, rather than simply removed, to inform users of the snapshot-before-destroy alternative (encouraging proactive config changes to avoid accidental data deletion events). Terraform updates to add snapshot-before-destroy are quick and in-place.
3. While the auto_delete=false does not impact deletion behavior due to the google_compute_disk's separate destroy lifecycle, it is important to enable the snapshot-before-destroy feature. If the disk is deleted in connection with the VM instance (i.e., auto_delete=true), then the snapshot does not trigger. If the disk is deleted during it's own dedicated destroy lifecycle (i.e., auto_delete=false), then the snapshot does trigger. So the auto_delete has no impact on module deletion behavior, but it needs to be hard-coded as auto_delete=false for snapshot enablement.

Manually tested a variety of scenarios with Terraform v1.11.3 and Packer v1.12.0

• ✅ Fork your PR branch from the Toolkit "develop" branch (not main)
• ✅ Test all changes with pre-commit in a local branch #
• ✅ Confirm that "make tests" passes all tests
• ✅ Add or modify unit tests to cover code changes
• ✅ Ensure that unit test coverage remains above 80%
• ✅ Update all applicable documentation
• ✅ Follow Cluster Toolkit Contribution guidelines #