Dashboard workflow

The dashboard is designed for a simple loop:

1. Upload inputs and create a dataset.
2. Queue a run and monitor status.
3. Inspect results/artefacts.
4. Run optimisation.

Main sections

• Data: upload inputs and create datasets; enqueue a dataset to start a run.
• Training jobs: create runs, view run history, inspect artefacts/logs, delete/cancel runs.
• Results: browse outputs for a selected run (plots + key tables).
• Optimiser: run single-period optimisation against a completed model run.
• Agent: ask questions grounded in results/docs/logs (requires LLM config).
• API keys: create/revoke API keys for v2 endpoints.
• System status: health check for DB/Redis/Storage/LLM.

TODO: Scenario planner route is under active development; document once enabled.