Parameterize collection endpoint tests across media types, consolidate test infrastructure

[mforce]

What to build

Parameterize the collection endpoint test suites so all three media types get identical coverage, and consolidate the copy-pasted test infrastructure.

Today the movies/music/games endpoint test files triplicate 16 identically-named scenarios (auth, ownership, CRUD round-trips, tag handling, validation), while 11 further behaviors — currency normalization, boundary ratings, tag replacement semantics, cover URL handling, owner-scoped tags, and more — are verified for movies only even though the production code paths are byte-identical across the three types. The suite also re-declares response DTOs and Sample() factories per file, hand-rolls the same HTTP stub handler in four places, and boots a full web host per test (186 times), each boot replaying all migrations.

Convert the shared scenarios into one parameterized (or shared-base) suite that runs per media type, extract a single test-support module (DTOs, sample factories, stub-handler family, class-level host fixture), and align unit-level tests with the endpoint tests on the migration-based schema path so a model/migration divergence is visible to the whole suite.

This is the safety net for the collection-endpoint consolidation — it must land first.

Acceptance criteria

• The shared CRUD/auth/ownership/validation/tag scenarios run against movies, music, and games from one definition
• The behaviors currently verified only for movies are covered for music and games
• Response DTOs, sample factories, and HTTP stub handlers are declared once in shared test support
• Unit-level and endpoint-level tests build their schema the same way
• Suite wall-clock time is the same or better than before

Blocked by

None — can start immediately.

[mforce]

Shipped via PR #143 (merged). The parameterized collection-endpoint test suite is the safety net in place. StubHandler consolidation from acceptance criterion #3 is tracked in #144.