US: feature flags API with elevated error rate, delayed realtime destinations

We are fully caught up and operational without realtime destination delays.

We're monitoring while the rollback is progressing - feature flags returned to normal values and destinations are still catching up.

The rollback is progressing and performance and error rates have recovered.

Due to this faulty config change, we accumulated 2-3 minutes of delayed realtime destinations that are currently catching up.

We identified a faulty change that we are currently rolling back. We're monitoring and seeing signs of recovery.

We've spotted that something has gone wrong and observing an elevated error rate with feature flags. We're currently investigating the issue, and will provide an update soon.