CI: Add GitLab Runner Tags for Very High CPU with RAMFS Support

What

This MR introduces new GitLab runner tags gcp_very_high_cpu_ramfs_dev and gcp_very_high_cpu_ramfs to support CI jobs that require:

• High CPU resources (64 vCPUs via n2d-highcpu-64 instances)
• RAM-backed filesystem for the /builds directory using tmpfs
• Improved I/O performance for build-intensive operations

Changes Made

• Added runner_tezos_very_high_cpu_ramfs_amd64_tags configuration in locals.tf
• Configured tags for both development and production environments
• Implemented corresponding node pool node_pool_amd64_very_high_cpu_ramfs with appropriate machine types and scaling limits

Technical Details

The new runner configuration:

• Uses n2d-highcpu-64 instances (64 vCPUs, 64GB RAM)
• Mounts /builds directory on RAMFS for faster I/O operations
• Supports auto-scaling from 0 to 64 instances
• Runs on spot instances for cost optimization

About arm64 CI jobs

This change clarifies our current infrastructure setup, currently, ARM64 workloads already utilize RAMFS storage implicitly and run on a separate pool with different specifications (lower CPU/memory compared to the new pool being introduced, and specifically on ARM64 architecture nodes).

By introducing the explicit storage concept in CIAO, I needed to make previously implicit configurations explicit, this is why you'll see ~storage:Ramfs added to ARM64 jobs - it reflects the existing behavior rather than introducing new functionality.

Note that AMD64 jobs remain unchanged in this MR. The scope here is limited to adding the storage tag infrastructure to CIAO without actively utilizing it, modifications to AMD64 job configurations will be addressed in subsequent MRs.

Usage

CI jobs can now use these tags to access the high-performance RAMFS-enabled runners: