[🚀] Create a better approach for retry logic

[Nek-12]

• We can try using coroutines's selectors for better stability.
• Additionally, we should somehow handle the case where a coroutine launches a child coroutine.
  I believe for that we'll need to write a dsl for creating child PipelineContext instances?
• Also we should add a selector lambda to time-out only specific intents, and if not chosen, skip the overhead.