Currently, we update the value of intensities to newValue after calling healthCheck(). If the timing is right, this can lead to a scenario where the resulting health check is a no-op (because the observer is seen as disabled) even if the check returns a failed status and the observer was just enabled. Further, another health check won't be scheduled and this observer will remain effectively disabled. To fix, we should call healthCheck() after updating to new intensity values so that the check uses the new values.