Informer initial start exception handling

[shawkins]

Is your enhancement related to a problem? Please describe

The startup of an informer is treated differently than exception handling once it is running. On startup any problem with the initial list or watch will be treated as a hard failure, while later on it will keep retrying. See also #4365

Describe the solution you'd like

To conform with the go client we should let the informer simply keep retrying from the start. If the user wants to know if it has ever started, there's the hasSynced method, or we can add another CompletableFuture similar to stopped.

We could apply this change to Informers stated by runnableInformer without much of a risk of compatibility issues. All other paths though the user is probably expecting an informer that has already synced to be returned.

Describe alternatives you've considered

No response

Additional context

No response