[OpenSTEF 4.0] Optimize tutorial notebooks for CI execution

[egordm]

Context

PR #885 integrates myst-nb notebook rendering infrastructure. Notebooks currently render as static code (no outputs) because execution exceeds the 120s timeout.

Parent: #883

Goal

Optimize tutorials so they execute within 120s, then flip nb_execution_mode from "off" to "cache" so rendered docs show actual outputs.

Tasks

Notebooks that can be made fast

• forecasting_with_workflow_presets.py: Reduce dataset to ~14 days, use minimal model params (n_estimators=10)
• hyperparameter_tuning_with_optuna.py: Use create_synthetic_forecasting_dataset(length=timedelta(days=14)), reduce to 3 Optuna trials

Notebooks that cannot be made fast

• backtesting_openstef_with_beam.py: Move to examples/benchmarks/ (not rendered in docs). Create a simpler "backtesting quickstart" tutorial that demonstrates the API with 1-2 horizons and a tiny dataset

Rendering quality

• Switch all tutorial plotting to PNG-based matplotlib backend (%matplotlib inline or matplotlib.use("agg")) — avoid interactive HTML plots (plotly/bokeh) which produce large outputs and render poorly in static docs
• Add disclaimer at top of each tutorial: "Uses minimal data for fast execution. See examples/benchmarks/ for production-scale runs."

Infra

• Enable nb_execution_mode = "cache" in docs/source/conf.py
• Add .jupyter_cache/ to GitHub Actions cache for faster CI builds
• Set timeout to 120s (already configured)
• Verify poe docs builds with outputs in < 3 minutes total

Constraints

• Each notebook must complete execution within 120 seconds
• No interactive HTML widgets in output (PNG plots only)
• create_synthetic_forecasting_dataset is the preferred data source for tutorials
• Backtesting benchmark stays in examples/benchmarks/ — NOT in docs toctree

Related

• Blocked by: docs: integrate myst-nb notebook rendering infrastructure (#884) #885 (infrastructure)
• Parent: [OpenSTEF 4.0 Epic] Documentation #883 (Documentation Epic)
• Sibling: [OpenSTEF 4.0] Add support for jupytext and myst-nb for converting jupyter notebooks to docs #884 (Notebook infrastructure plan)