g5k: add automatic retry when synchronizing jobs
Currently we only make one attempt to synchronise jobs.
- Either that succeeds and you are happy (and the jobs start immediately)
- Either that fails and you'll have to make a reservation manually (and you aren't happy)
We can fix the latter by adding a retry logic in the code that will find a slot for you in the future. A good place for that is in: https://gitlab.inria.fr/discovery/enoslib/-/blob/a1583dfd2faa59a0e651496b645f7f8603a5871d/enoslib/infra/enos_g5k/g5k_api_utils.py#L593-651
A first implementation can test slots every 5 minutes in the next XX (12?) hours.