Python ↔ Rust feature parity

The name-intelligence feature set is implemented in both languages. This is the authoritative parity matrix.

Real name ground truth (`faker2.naming.realnames` / `real-names` feature)

Capability	Python	Rust
Infer gender from real data	`realnames.infer_gender(name, country)`	`Faker::infer_gender_real(name, country)`
Frequency-weighted first name	`realnames.first_name(country, gender)`	`Faker::first_name_real(country, gender)`
Gender-preserving replacement	`realnames.first_name_like(name, country)`	`Faker::first_name_like_real(name, country)`
Country detection	`realnames.detect_country(name, top)`	`Faker::detect_country(name, top)`
Homophones (metaphone/ipa/levenshtein/balanced)	`realnames.homophones(...)`	`Faker::homophones(...)`
Per-country balanced weights	`data/balanced_params.json`	same file (via `serde_json`)
List countries	`realnames.available_countries()`	`Faker::available_countries()`

Capability	Python	Rust
Infer gender (bundled lists)	`gender.infer_gender(name, locale)`	`Faker::infer_gender(name, Locale)`
Gender-preserving replacement	`gender.first_name_like(name, locale)`	`Faker::first_name_like(name, Locale)`
Gendered full name	`gender.full_name(locale, gender)`	`Faker::name_of(gender, Locale)`
pluralize / singularize	`grammar.pluralize` / `singularize`	`grammar::pluralize` / `singularize`
indefinite_article / agree / is_are	`grammar.*`	`grammar::*`

Bundled-locale breadth: Python gender.* works over every faker locale (Faker(locale)), Rust gender.rs ships only Locale::{EnUS, FrFR} embedded lists. For broad coverage use the realnames path in both — it spans the same 139 countries either side.
Gender code case: realnames uses lowercase "m"/"f"; the bundled gender module uses "M"/"F". full_name accepts either.
Seeding: both are deterministic within a language but not byte-identical across languages (different PRNGs).