Revisit the design of scheduler of coordinator

Currently for scheduled jobs, we could not get the job id and `submit_dataloading_job` function is called. Could we assign a unique job id when the API is called, and provide it as a parameter to the real loading service?