Should curl requests have a shorter timeout?

If there's a problem with non-responsive SCIM server it seems a run can take a very long time.

Perhaps there's a timeout setting in curl that could/should be modified.

We should perhaps also give up after a few failed attempts? If so we still need to handle remaining objects properly so they're transferred to the new cache-file so we'll try again next time.