[Feature flag] Rollout of `duo_evaluation_ready`

## How this feature flag is used When [evaluation runner](https://gitlab.com/gitlab-org/modelops/ai-model-validation-and-research/ai-evaluation/evaluation-runner) kicks off an evaluation pipeline which launches a remote GDK instance, it executes `wait-for-instance` job to make sure the instance is ready for evaluation, [example](https://gitlab.com/gitlab-org/modelops/ai-model-validation-and-research/ai-evaluation/evaluation-runner/-/pipelines/1552191316). We're currently checking [the HTTP status from GitLab-Rails url](https://gitlab.com/gitlab-org/modelops/ai-model-validation-and-research/ai-evaluation/evaluation-runner/-/blob/main/scripts/wait-for-instance.sh?ref_type=heads), however this technically returns 200 even before [the post-deployment script](https://gitlab.com/gitlab-org/modelops/ai-model-validation-and-research/ai-evaluation/evaluation-runner/-/blob/main/dockerfiles/gdk/scripts/setup.rb?ref_type=heads) is complete. We enable the `duo_evaluation_ready` feature flag at the end of the post-deployment script, and change the `wait-for-instance` job to request to `http://gdk.test:3000/api/v4/features` endpoint to get the state of the FF. So that the evaluation won't start until the GDK is ready.

issue