[what happened] Dosing automation base's has a race condition that doesn't ended the job properly.

Suppose I have a long-running execute, ex:

def execute(...):
        from time import sleep
        from pioreactor.actions.pump import remove_waste, add_media
        for i in range(10):
            add_media(duration=1.0, unit=self.unit, experiment=self.experiment)
            sleep(5)
            remove_waste(duration=3.0, unit=self.unit, experiment=self.experiment)
        return

If I try to stop the dosing_controller job (from the UI) mid-execute, the pumping doesn't stop, and eventually LOST on the controller is thrown. The pumping doesn't stop because the automation_job has a thead.join() in it's on_disconnected.

Users should put in logic to check for state here, but does this also happen for our execute_io_action?

Start at automation_job.on_disconnected and its thread.join() behavior, then compare it with execute_io_action. Reproduce the issue by stopping a long-running execute from the UI; done means the pumping stops and the job ends without the controller reporting LOST.