Timer execution has unexpected results when using short timers (less than 1min) and Job Executor has default settings

See: https://forum.camunda.org/t/email-reminders-using-two-boundary-event-timers/8430/17?u=stephenott

When a timer is a "short timer", shorter than the job executor's wait settings, the executor seems to have a race condition occurring where it is executing the same job hundreds of times. This only occurs on the first run of the instance of a new deployment (tested with Rest API). On the second run of a instance, the issue does not occur. If you deploy the same definition a second time, the issue occurs again for the first instance.