notebooks/Synthea.ipynb: rewrite as a real synthlab tutorial

[bschilder]

Summary

The Synthea notebook on `main` was byte-identical to `bschilder/AoU/notebooks/Synthea.ipynb` — it still imported `AoU.phenome.synthea.*` and never exercised any synthlab code. This PR replaces it with an end-to-end tutorial against synthlab's public API.

Walks through:

1. `SyntheaRunner()` auto-downloads the Synthea JAR
2. `SyntheaConfig(population_size=10, exporter_flags={"exporter.csv.export": "true", "exporter.fhir.export": "false"})` enables CSV output (Synthea defaults to FHIR JSON; the OMOP converter needs CSV)
3. `runner.run()` → 18 CSV files
4. `convert_synthea_to_omop(cdm_version="5.4", output_format="parquet")` → 10 OMOP tables (person, condition_occurrence, drug_exposure, measurement, visit_occurrence, …)
5. `pandas` previews of the resulting OMOP frames

Render status

• 6 code cells, 6 with outputs, 0 errors
• person table preview shows 14 real synthetic patients (10 living + 4 historical from the 10-pop seed)

Notes for reviewer

• The notebook `%load_ext autoreload` cell appends openjdk@21 to `PATH` if the homebrew install is present; this is the only platform-specific bit and only kicks in on macOS. Linux/CI: pre-install Java.

Test plan

• Open on github.com and confirm outputs render
• `jupyter nbconvert --to notebook --execute --inplace notebooks/Synthea.ipynb` reproduces locally (requires Java)

🤖 Generated with Claude Code