Extend op-service LifecycleCmd usage to all long-running Go services

[protolambda]

Op-node uses cliapp.LifecycleCmd to handle startup/shutdown/force-shutdown, while ensuring resources are closed and any shutdown errors are propagated properly.

Previous context-based shutdown patterns encouraged bad code that shuts down like defer thing.Close() without propagating the error, or even returning before go-routines of the service fully shutdown, since context closure triggers shutdown, but does not await the results.

Services to update:

• [x] op-batcher
• [x] op-proposer
• [ ] endpoint-monitor
• [x] indexer
• [ ] op-wheel (engine auto sub-command is long-running, other sub-commands are not, may require some more work)
• [ ] op-bootnode
• [ ] op-heartbeat
• [x] op-challenger

[mslipper]

@protolambda are you still working on this?

[protolambda]

@mslipper an external contributor implemented the op-proposer changes last week. op-batcher and op-node are already done. Op-challenger remains I think.

[protolambda]

The indexer is also done. And I think op-bootnode, op-heartbeat and op-wheel are nice to have; they work relatively well as-is since they don't have a lot of resources to start/close, but I may be able to implement the changes there soon now that I'm back from devconnect.