Expose promptInjectionProtectionLevel field via Graphql

What does this MR do and why?

Expose the field introduced in Add prompt_injection_protetion_level backend field (!216829 - merged). UI/UX for it is introduced in feat: Add support for multiple prompt scanners (gitlab-org/modelops/applied-ml/code-suggestions/ai-assist!4149 - merged)

Related issue: https://gitlab.com/gitlab-org/gitlab/-/issues/584247+

Testing

{
  group(fullPath: "gitlab-org") {
    aiSettings {
      promptInjectionProtectionLevel
    }
  }
}