While #33736 added the ability to bill partial stream usage spend for requests that disconnect mid-stream, I noticed that AWS Bedrock (via checking CUR logs) charges the full token usage for this scenario. Is it possible that we add a flag in the `config.yaml` file where we can choose whether to let the stream bill on the full usage amount (i.e., drain the upstream on client disconnect mid-stream) instead of just the partial usage? _Originally posted by @nuernber in https://github.com/BerriAI/litellm/issues/33736#issuecomment-5025961420_