Add gcp-cloudrun-utilization-health CodeBundle

[rw-codebundle-agent]

Closes #155

Summary

Adds the gcp-cloudrun-utilization-health CodeBundle to rw-cli-codecollection. It monitors resource utilization and scaling configuration of GCP Cloud Run services per project, flagging over-provisioned (under-utilized), over-utilized, and improperly scaled services, and captures a utilization/scaling report for LLM-based cost and sizing review.

Tasks

• Check Cloud Run Service CPU Utilization — check_cpu_utilization.sh, flags services at/above CPU threshold (sev 3), data:metrics
• Check Cloud Run Service Memory Utilization — check_memory_utilization.sh, flags OOM risk (sev 3), data:metrics
• Check Request Concurrency and Instance Scaling — check_concurrency_scaling.sh, flags unbounded max instances (sev 3), low concurrency and idle-warming min-instances (sev 2), data:config
• Identify Under-Utilized Services — find_underutilized_services.sh, flags idle/over-provisioned services (sev 2), data:metrics
• Report Utilization and Scaling Configuration — capture_utilization_report.sh, captures metrics+config for LLM review (no issues)

SLI

In-repo sli.robot produces a binary 0-1 health score across CPU, memory, and scaling dimensions (pushed as sub-metrics with raw issue counts) with the aggregate pushed as the primary metric. Generation rules include - type: sli and a *-sli.yaml template is provided.

Verification

• Scorer: 104/104 (PASS), no critical failures
• Bash scripts validated (bash -n) and logic unit-tested against mocked gcloud/Cloud Monitoring responses
• Test infra (.test/): Taskfile + Terraform fixtures

Deviations from the design spec

• Resource type: spec lists gcp_cloud_run_service; used gcp_run_services (the CloudQuery table name used by the collection indexer) so discovery matches.
• CPU task scope: over-utilization is handled by the CPU task and under-utilization by the dedicated under-utilized task to avoid duplicate issue generation, matching the spec's expected test-scenario issue counts.