Adding Support for updating Availability Zones

[bamdadd]

This pull request refactors the Terraform configuration to use for_each instead of count for resources that are created per availability zone. This change improves resource management by making it easier to add or remove availability zones without destroying existing resources. Outputs are also updated to use explicit mapping by availability zone, and a new integration test is added to verify that adding an availability zone does not destroy existing resources.

Resource management improvements:

• Refactored all per-availability-zone resources in public_subnets.tf, private_subnets.tf, and nat.tf to use for_each with toset(var.availability_zones) instead of count, ensuring stable resource addressing and preventing unnecessary resource destruction when availability zones are changed. [1] [2] [3]
• Updated resource references and tags to use each.value (the AZ name) for clearer resource identification and mapping. [1] [2] [3]

Output consistency:

• Modified all outputs in outputs.tf to explicitly map resource IDs and CIDR blocks by availability zone, ensuring outputs remain stable and predictable when AZs are added or removed.

Testing and validation:

• Added a new integration test spec/integration/availability_zone_addition_spec.rb that verifies adding an availability zone results only in new resources being created, with no destruction of existing subnets, NAT gateways, or EIPs.

[jonassvalin]

Breaking change so will need a major version upgrade