Job control with working model and hazard stages
Hazard Stage
Implements the hazard stage with an OpenQuake client for the OpenQuake 2 rest API.
Job Control
This MR improves job control by
- Making return codes from workers more informative and correctly handling them on the client side
- Correctly evaluating job statuses and and reflecting them in the gui
- Persisting job statuses to the database
Tested with the remote ollinger worker on ramsiswin.ethz.ch and and OpenQuake running on localhost.
ToDo
- Adapt the other workers (use the same code as for ollinger)
- Indroduce timeouts / max. retries for the clients (model, hazard, risk)
- Pass results between stages