refactor: Move synchronisation logic to LifecycleManger

[morgsmccauley]

This PR moves the synchronisation logic within ExecutorsHandler/BlockStreamsHandler in to LifecycleManager. The motive behind this is to provide greater awareness/control over the individual sync tasks (managing unhealthy, reconfiguring, etc.) to LifecycleManger.

I plan to build on top of this in my next PR, and have LifecycleManager count how many times an unhealthy stream/executor was restarted, and move the Indexer to Suspended if it goes above a configured threshold. This refactor makes "counting" actually possible.

Wanted to land the refactor PR first to make review easier.

[darunrs]

Can we retain the restart timeout until the retry configuration is in place? The timeout of retries originally helped in two cases: Block streams which would restart over incorrect stale checks, and executors which constantly crash. The former is partially remediated due to the stale check being longer. But the latter is still an issue. Basically, I believe these indexers make some async call which fails, and it escapes our try catches for some reason. As a result, it crashes. This dumps logs into Runner and is also unnecessarily taxing on the system as restarts involve block prefetches. It also messes with the metrics. I have not gotten around to fixing any of those issues yet. Keeping in the timeout on retries at least mitigates the issue until we either have retry policies in place.

[morgsmccauley]

Ah sorry, I meant to add the timeout back in but forgot - will do that now!