Describe the solution you'd like
If the workflow task failure is due to gRPC message too large, set GrpcMessageTooLarge as failure_reason on temporal_workflow_task_execution_failed. This should not overly expand cardinality on users with properly operating workers/workflows.
Also while doing this, please confirm this "gRPC message too large" determining condition/code works on cloud as expected instead of just self-hosted/dev-server.
Per-SDK Tickets