[Engine] Ray actor scaler during runtime

[Superskyyy]

Data could come in at peak throughout the day, and the stream size in Redis may grow indefinitely, outpacing ray-consuming speed and eventually breaking the application. It's imperative to provide scaling for actors (internal operator), then the autoscaler from Ray will handle the rest (pod scaling)

This issue tracks the creation of a new Ray Actor that manages the creation and destruction of actors during runtime.

[Superskyyy]

Closing in favor of our movement to Flink (built-in support is provided).