Document ETL CSV usage across the repository

[GVCUTV]

Motivation

• Provide a single, code-driven reference that maps every ETL-produced CSV to its producer, schema, semantic role, and downstream consumers.
• Surface mismatches and fragilities between ETL outputs and simulation/validation consumers to aid auditing, validation, and refactoring.

Description

• Add docs/analysis/etl_csv_usage_analysis.md describing ETL dataflow (download → clean/merge → enrich → fit), one section per CSV (producer, inferred schema, semantics, consumers), and auxiliary exploration outputs.
• Enumerate direct consumers in simulation/ and validation/ (notably tickets_prs_merged.csv and fit_summary.csv), orphan/legacy CSVs, and observed schema/selection drifts (e.g., fit_summary.csv generator differences and feedback/capacity column dependencies).
• Summarize critical dependencies, fragilities, and potential sources of inconsistency to guide next steps for alignment and testing.

Testing

• Documentation-only change; no automated tests were executed as part of this update.

---

Codex Task