estuary-cdk: Add restart_interval override

[jshearer]

Description:

This allows connectors to override their target lifetime. The problem is that some connectors may have tasks that run for longer than the restart interval. Since the restart signal is soft, i.e all tasks are allowed to run to completion before the connector exits (but no new tasks will be started once the restart signal has been triggered), tasks running longer than the restart interval could effectively starve the shorter-running tasks.

Note: This does feel like kind of a band-aid fix, but I'm not thinking of a better option right now. Needing to restart at all feels like a bit of a smell... so maybe this should be paired with some logic to just never (voluntarily, we should still handle external exit requests properly) restart if not needed?

Also, ideally this would be hidden under an Advanced section, but I wasn't confident enough in my knowledge of Python OOP best practices to define a generic/abstract base BaseAdvanced class, and then have the abstract BaseEndpointConfig include it, while also allowing both to remain extendable by connector authors. If you feel strongly that it should be in Advanced then I'll go and figure this part out, if not I'm happy to leave it as is for now.

---

This change is 

[jgraettinger]

I believe this is not desired, especially given the evolution of the NetSuite connector, and can be closed out.