Normalize and reorganize configurations of Onyxia services

[avouacr]

To improve user experience, I think we should :

• normalize configurations of Onyxia services as much as possible. For instance, the RStudio service has a "R" configuration to choose the image and a "Service" configuration to activate Spark-history. On the other hand, Jupyter only has the "Service" one for both uses. This inconsistency can cause issues, for instance : in the "R" configuration, one can use a custom image, but not choose the pull policy, so that the custom image will never be updated if a previous version is already present.
• reorganize the order of configurations to put first the ones that are most likely to be changed by users. Right now, the first one is S3, which is rarely modified. Here is a suggested order for Jupyter/RStudio (based on my use..) :
  R/Jupyter - Init - Git - Kubernetes - Resources - Persistence - Security - Discovery - all the rest

So, in terms of tasks :

• Normalize configuration schemas as much as possible between services
• Think about the best ordering for all possible configurations, based on how most users configure their services
• Add an "order" object in the schemas of the helm charts to reflect this ordering
• Retrieve this parameter in the UI to implement the ordering