fix(spartan): drop broken Terraform validations on RPC ingress vars

[AztecBot]

The Nightly Spartan Scenario CI broke on v5.0.0-nightly.20260425 (run). Both ci-network-scenario (1) and (2) jobs died inside terraform plan for deploy-aztec-infra after about 5 seconds.

Root cause

The RPC_INGRESS_LOG_SAMPLE_RATE variable added in #21591 (chore: setup CloudDNS and CloudArmor) had this validation:

condition = var.RPC_INGRESS_LOG_SAMPLE_RATE == null || (var.RPC_INGRESS_LOG_SAMPLE_RATE >= 0 && var.RPC_INGRESS_LOG_SAMPLE_RATE <= 1)

|| does not short-circuit inside a Terraform validation condition — Terraform eagerly evaluates the var.X >= 0 subexpression, and null >= 0 errors with argument must not be null, aborting the plan. next-scenario does not set this var, so deploy_network.sh writes the default literal null into the generated terraform.tfvars, which trips the validation.

Yesterday's nightly tag (v5.0.0-nightly.20260424 at 848af53aa) was cut before that commit landed.

Fix

Drop the validation blocks on the variables that #21591 added: RPC_INGRESS_LOG_SAMPLE_RATE and RPC_INGRESS_SESSION_AFFINITY. Both are forwarded into the GCE BackendConfig, which validates the values server-side; the Terraform-level checks were duplicating that validation and the LOG_SAMPLE_RATE one was outright broken.

Full analysis (failure logs, reproduction, etc.): https://gist.github.com/AztecBot/14f1172ad743662595c1bdc630958d99