Improve thread safety to avoid race conditions during shutdown.

[alexcams]

This PR implements thread safety mechanisms during application shutdown and integration tests to avoid transient errors such as these.

Plugin change:

As described in #67 there was a race condition where the consumer tried to ack a message when the channel was closed, raising an ChannelAlreadyClosed exception. This changes include a sync between the plugin shutdown and the consumer thread to avoid acking messages when the channel has been closed.

Integration tests improvements:

Previously, the consumer thread created during each test startup was not stopped gracefully at test closure. This could cause errors , as the consumer thread might attempt to interact with a channel that had already been closed after the test finished.

To address the mutex synchronization error during exchange deletion, the deletion now occurs only after the consumer thread has been gracefully stopped using join function.

Another improvement is the deterministic wait for the channel to have consumers, replacing the previous sleep 4. This change makes test execution faster and more reliable, reducing the wait time from 4 seconds to 1 second per test.

Thanks for contributing to Logstash! If you haven't already signed our CLA, here's a handy link: https://www.elastic.co/contributor-agreement/

[kaisecheng]

LGTM. Thanks for improving and speeding up the tests