#113 chore: moved init stuff to DI

---

Summary by cubic

Moved service startup and lifecycle management into DI and refactored workers/services into stateless handlers. Added Redis-backed repositories and stronger typing to improve safety, scalability, and multi-instance behavior.

• Refactors

  • Removed LifecycleEnabled and lifecycle-heavy services; converted Coordinator, K8sWorker, PodMonitor, ResultProcessor, Notification, DLQ, Saga, SSE bridge, and EventBus into stateless handlers.
  • DI now provides ready AIOKafkaProducer/Consumer; workers iterate the consumer directly.
  • Deleted QueueManager, ResourceManager, EventStoreConsumer, idempotency manager/middleware, and SSE health endpoint; simplified dishka_lifespan and providers.
  • Adopted strong types: KafkaTopic for DLQ topic fields, QueuePriority for execution priority, and required user_id in EventMetadata; updated tests to match.

• New Features

  • Added Redis-backed repositories: IdempotencyRepository (SET NX), PodStateRepository, UserLimitRepository.
  • Scheduling limits and pod tracking moved to Redis for atomic ops and multi-instance safety.
  • EventBus exposed as a stateless service via DI for cross-instance pub/sub.

Summary by CodeRabbit

• Bug Fixes

  • Enhanced system stability through internal architecture improvements.

• Chores

  • Refactored service infrastructure for improved performance and maintainability.
  • Implemented per-user execution limits (maximum 100 concurrent executions by default).

_{✏️ Tip: You can customize this high-level summary in your review settings.}