Add regression tests for locale default hour cycle (Fixes #594)

[afrdbaig7]

Fixes #594

This PR adds regression tests to make sure datetime formatting falls back
to the correct default hour cycle based on CLDR region data when no
explicit -u-hc- preference is provided.
I added both runtime and datagen tests so that the behavior is covered
at formatting time as well as at the data level.

Runtime tests (simple_test.rs):

• Check default behavior:
  • en → h12
  • fr, ja, en-GB → h23
• Verify that explicit -u-hc- overrides still take priority
• Add a midnight edge case to ensure formatting is correct

Datagen test (semantic_skeletons/tests.rs):

• Validate that preferred_hour_cycle() returns the expected
  CoarseHourCycle:
  • en → H11H12
  • fr, ja, en-ZA → H2
    Note: The runtime tests require feature = "experimental" for
    components::Bag access.
    These tests should help prevent regressions and make the expected
    locale-based hour cycle behavior more clearly covered.

[Copilot]

Pull request overview

This PR adds comprehensive regression tests for issue #594, ensuring that datetime formatting correctly falls back to locale-specific default hour cycles based on CLDR region data when no explicit -u-hc- preference is provided.

Changes:

• Added datagen test to validate that preferred_hour_cycle() correctly infers hour cycles from CLDR data
• Added runtime test to verify that datetime formatting uses the correct default hour cycle for various locales
• Included tests for explicit -u-hc- overrides and midnight edge cases

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
provider/source/src/datetime/semantic_skeletons/tests.rs	Adds test_preferred_hour_cycle_by_locale() to validate datagen pipeline correctly infers locale-specific hour cycles from CLDR patterns
components/datetime/tests/simple_test.rs	Adds test_locale_default_hour_cycle() to verify runtime formatting behavior with default and explicit hour cycle preferences

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Consider importing Gregorian from icu_calendar at the top of the file and using the shorter Gregorian instead of the fully qualified icu_calendar::Gregorian. This would be consistent with other test files like resolved_components.rs that import and use Gregorian directly.

[sffc]

I think we have tests for this already, but adding additional tests specifically for the components::Bag behavior as integration tests is harmless and directly fixes the issue.

[afrdbaig7]

I think we have tests for this already, but adding additional tests specifically for the components::Bag behavior as integration tests is harmless and directly fixes the issue.

Thanks for the review
Yes, the main addition here is the components::Bag integration test in simple_test.rs. It structurally checks the resolved hour_cycle for different locales (for example, en → H12 and fr/ja/en-GB → H23). It also covers explicit -u-hc- overrides to make sure they still take precedence, along with a midnight edge case to ensure formatting behaves correctly.
Additionally, I added a datagen-level test in semantic_skeletons/tests.rs that verifies preferred_hour_cycle() directly against the CLDR source data. This acts as an extra safety check so that both the data layer and the runtime behavior stay aligned.

[sffc]

Please fix the merge conflict