Potential race condition while closing Worker

As observed in a CI run (but very hard to reproduce, this run somehow reused a CI container from a previous run which might have polluted the environment): https://github.com/scylladb/scylla-cdc-java/runs/3477278901

The issue seems to arise when two threads concurrently run Worker::stop() and Worker::makeCallable() "continuation":

Thread A:
https://github.com/scylladb/scylla-cdc-java/blob/ab828eb0305171ffa48d6114dc73aaba0257ffdc/scylla-cdc-base/src/main/java/com/scylladb/cdc/model/worker/Worker.java#L153-L156

Thread B:
https://github.com/scylladb/scylla-cdc-java/blob/ab828eb0305171ffa48d6114dc73aaba0257ffdc/scylla-cdc-base/src/main/java/com/scylladb/cdc/model/worker/Worker.java#L114-L123

1. Thread B on line 118 checks shouldStop() and proceeds to go to else if body.
2. Thread A runs line 154 and 155, setting shouldStop=true and shutting down the executor service.
3. Thread B on line 119 tries to submit a task to executor service, which ends with java.util.concurrent.RejectedExecutionException as the executor service is now terminated.

Start in scylla-cdc-base/src/main/java/com/scylladb/cdc/model/worker/Worker.java at the referenced lines in stop() and makeCallable(). Trace the concurrent executor shutdown and task submission paths, then verify that concurrent calls do not produce RejectedExecutionException while closing the Worker.