Retry on timeouts and increase timeout when scanning OBS
For some of the longer running jobs it is sufficient for the service they interface with to be marginally slower than usual to trigger a timeout, either in the overall job execution time on on specific requests from the tools, causing the whole pipeline to fail.
In those case we can afford to retry the job to see if things improve.
In addition the job retrieving data from OBS often reaches the default 1h timeout, so let's be more lenient there.
Edited by Emanuele Aina