feat(notebooks): add 05_impact_reporting stakeholder template

[franchaise]

Summary

• Adds notebooks/05_impact_reporting.ipynb — end-to-end impact reporting template that combines analytics.carbon.estimate_carbon, analytics.statistics.compute_trend, and analytics.reporting.generate_report for a configurable region/period.
• Pulls in validation metrics from outputs/validation/benchmark_report.json (produced by the validation notebook in feat(notebooks): add 04_model_validation benchmarking notebook #48) and falls back to a baseline when not present.
• Produces both the structured JSON used by the /api/reports endpoint and a stakeholder-ready Markdown narrative under outputs/reports/.
• Region / bbox / period / forest_type are top-level constants so the same notebook re-runs for Amazon, Congo, or Southeast Asia by swapping a few lines.

Why

Sprint deliverable: "Create 05_impact_reporting.ipynb — template for regional impact reports." Closes the carbon-analytics loop: training (#46) -> analysis (#47) -> validation (#48) -> reporting (this).

Test plan

• Valid notebook JSON.
• Runs top-to-bottom on synthetic data via papermill without errors.
• Output Markdown opens cleanly in the GitHub renderer.

Notes for reviewers

• Designed to be scheduled quarterly via papermill once the analytics database is wired up.
• The narrative output is intentionally factual / non-promotional; the prose-smoothing pass is delegated to climatevision.reports.llm_reporter (Linda's feat(reports): add LLM-backed impact report generator #38) once both PRs land.

[femi23]

Data contract matches the /api/reports response shape we're serving in #42, so this notebook can be used as the reference implementation for what the API should return for a region/period query. The Markdown narrative is factual / non-promotional which is the right tone for the NGOs we're delivering to.

Approved.