Cover possible processor re-use with tests

[rdner]

Describe the enhancement:

We caught a high-impact bug recently #34761 and we need to introduce some testing in order to prevent this bug from reappearing. Or even to make sure we've fixed all the occurrences.

It should be enough to:

• create a config with a processor
• configure a Filebeat pipeline
• create a pipeline client
• close it
• create another pipeline client and try to publish an event
• after Add SafeProcessor wrapper #34647 and before Stop re-using processors defined in the config #34761 this should return an error

also, before both changes calling Close twice would cause a panic on add_kubernetes_metadata.

The test should reproduce the issue on the old code before being accepted.

[elasticmachine]

Pinging @elastic/elastic-agent-data-plane (Team:Elastic-Agent-Data-Plane)

[cmacknz]

It would be good to pair this with #34792, to identify why the Beat becomes non-functional and generates a seemingly infinite number of processing errors when a processor is reused so we can introduce a more graceful failure mode if this happens again.