Ignore job start when it is already running.

[EmileSonneveld]

The code implies that this is already the case: https://github.com/Open-EO/openeo-python-driver/blob/master/openeo_driver/views.py#L1031
However, this job got started twice and ended with an error: cdse-j-26041409461344bf8bdc70a3cc58e368 (ESA-APEx/apex_algorithms#463)

In my tests where I launch the job twice, one ends with an error, but the run that ends successfully is used for the final status: j-2604141121594893930e8bd01393dc76

To reproduce, start a job, wait till it is in status running, and then start it again

job = openeo.rest.job.BatchJob("j-2604141121594893930e8bd01393dc76", connection)
job.start_and_wait()

[soxofaan]

The code implies that this is already the case

indeed, the underlying problem seems to be short-term inconsistencies when using elastic search as db backend