swh/rpc: adapt autoscalings based on prometheus metrics to use gunicorn metrics
- Adapt the prometheus request
- Add a factor to allow adapting the behavior based on the configuration of the service (threads/load peaks/criticity/...)
- Adapt the previous prometheus autoscaling configurations
- Add the support to prometheus autoscaling to the scheduler
Migration of other rpc services to the guninicorn based autoscaling will be progressively done in further commits