Backpressure exposure limits for Akka Streams WebSocket sink in UI workflows

The goal is to select a backpressure exposure strategy for an Akka Streams pipeline that emits high‑frequency events to a WebSocket sink serving a UI client, while keeping the UI responsive and avoiding client‑side overload.

Constraints include the need for the client to honor the chosen flow‑control mechanism, the trade‑off between latency introduced by backpressure and resource usage (buffer sizes), and the lack of a documented best practice that applies universally across UI scenarios. The analysis assumes Akka Streams 2.6+ where backpressure is enabled by default and can be applied at source, flow, or sink.

Which backpressure exposure mechanism—protocol‑level flow control (e.g., WebSocket pause/resume), adaptive chunk sizing, or application‑level acknowledgments—best minimizes latency without risking buffer overflow for high‑frequency WebSocket streams?
How does the chosen strategy affect perceived UI responsiveness when client technologies vary in their support for flow‑control signals?