Define settings to train Rasa models in Backoffice [v2]

[turnerrainer]

[!NOTE] This is a follow-up issue based on #385 The business requirements remain the same, but the technical solution is changed

Acceptance Criteria

GUI

• [ ] Based on current UI
• [ ] Manipulate Rasa training and testing settings via Backoffice GUI

DSL

• [ ] GET REST endpoint /training/trained to fetch data about previous model trainings in JSON from llm_trainings
• [ ] GET REST endpoint /training/settings to fetch current Rasa training settings in JSON from train_settings

  [!IMPORTANT] Current Dockerfile-based solution must be replaced with a Postgres-based one

• [ ] POST REST endpoint /training/settings to update Rasa training settings in JSON in train_settings
• [ ] Resql DSL to save updates regarding planned trainings