refactor: modular architecture with unified error handling and DashMap concurrency

- Break monolithic records.rs into commands/, output/, parsing/ modules
- Add lib.rs with shared types (AppError, Car, BestCarRecord, etc.)
- Replace Arc<Mutex<HashMap>> with DashMap for concurrent access
- Add unified AppError enum with thiserror
- Add configurable CLI paths (--my-cars, --cars-db, --times-dir, --tracks-db)
- Add serde deserialization for Car struct
- Add CSV escaping for fields with commas/quotes/newlines
- Fix track name splitting: split_once -> rsplit_once for _a handling
- Add integration tests, proptest property-based tests, and unit tests
- Add rustfmt.toml configuration
- Update documentation (API, ARCHITECTURE, DATA_STRUCTURES, README)

src/commands/best.rs

@@ -0,0 +1,19 @@
+//! Best and very_best command implementations.
+//!
+//! Generates best.csv (all cars best on at least one track) and
+//! very_best.csv (cars best on >= min_count tracks).
+
+use top_drives::AppError;
+
+use crate::commands::{load_my_cars_or_warn, run_analysis};
+use crate::output::{write_best_csv, write_very_best_csv};
+
+/// Executes the best command: generates best.csv and very_best.csv.
+pub async fn run_best(my_cars_path: &str, cars_db_path: &str, times_dir: &str, min_count: usize) -> Result<(), AppError> {
+    let my_data = load_my_cars_or_warn(my_cars_path)?;
+    let (_ctx, counts) = run_analysis(&my_data.owned, cars_db_path, times_dir).await?;
+    write_best_csv(&counts)?;
+    write_very_best_csv(&counts, min_count)?;
+    println!("Written: best.csv, very_best.csv");
+    Ok(())
+}