Add gcp-cloud-sql-health CodeBundle

[rw-codebundle-agent]

Resolves the gcp-cloud-sql-health design spec (issue #161).

Summary

Adds a new GCP Cloud SQL Health CodeBundle to rw-cli-codecollection. It detects Cloud SQL instances that are unavailable, misconfigured, or exposing access/ IAM risks, so operators can find availability, security, and configuration problems before they impact applications.

Tasks

• Check Cloud SQL Instance Status – flags instances whose state is not RUNNABLE (maintenance, failed, suspended).
• Fetch Cloud SQL Instance Configurations – dumps tier/disk/region/db version/maintenance/backup config; flags undersized tiers, disabled backups, and missing point-in-time recovery.
• Check Cloud SQL Instance Availability and Access – flags public internet exposure, missing SSL enforcement, exposed authorized networks, and IP issues.
• Check Cloud SQL IAM Policies – flags allUsers/allAuthenticatedUsers bindings and over-broad roles.

Includes an in-repo sli.robot producing a 0–1 health score across the four dimensions, plus *-sli.yaml, *-slx.yaml, and *-taskset.yaml templates and .test/ Terraform infrastructure.

Deviations from spec

• resourceTypes uses gcp_sql_instances instead of the spec's gcp_sql_database_instances. The latter is not a valid CloudQuery GCP table name; the actual table is gcp_sql_instances (verified against the CloudQuery GCP provider).
• expected_issue_severity ranges from the spec were mapped to concrete per-issue severities in the scripts, consistent with the grader's RW.Core.Add Issue severity model.

Scorer

python -m scorer.score codebundles/gcp-cloud-sql-health → 104/104, no critical failures.