[EPIC] MultibackendJobmanager improvement

[HansVRP]

In this EPIC we sketch out desired features for the MBJM.

To make the MBJB more performant, the downloading will be split of from track statuses. Apart from that we will make downloading optional and investigate further the need for threading of the download in case the download time is longer than the processing time.

Other requested features in terms of robustness include automatic retrying. Caution need to be taken to not rerun erroneous openEO jobs. As a start we will automatically rerun jobs which failed upon creation or starting.

Desired features:

• improve performance
• making the download optional
• automatic retry of jobs which failed during creation or start
• [SPIKE] investigate more convenient folder structure for downloaded results

[HansVRP]

@soxofaan