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

[AztecBot]

Fast-tracks the same fix as #22786 directly to next so the nightly next-net deploy stops failing while the merge-train/spartan PR (#22779) is still open.

Failure

Last two Deploy Next Net runs (https://github.com/AztecProtocol/aztec-packages/actions/runs/24948697881 and https://github.com/AztecProtocol/aztec-packages/actions/runs/24922876413, both at SHA a4ac021) failed during terraform plan for deploy-aztec-infra. The rollup-contracts deploy job succeeded (L1 contracts landed on Sepolia, registry 0x65ce7C5526Fe199e303B2Da7AfE96Bc9c066c7Eb), but no validators, provers, RPC, or bots were ever scheduled in the next-net namespace.

Root cause

The RPC_INGRESS_LOG_SAMPLE_RATE variable added in #21591 has 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-net.env does not set this var, so deploy_network.sh writes the default literal null into the generated terraform.tfvars, which trips the validation.

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. Identical to the diff in #22786.

ClaudeBox log: https://claudebox.work/s/929873c3e6b22a40?run=1

[AztecBot]

Automatically closing this stale claudebox draft PR (no updates for 5+ days). Re-open if still needed.