Add in-memory telemetry, Hilt DI, recommendations UI, fake repo and unit tests

[danoff]

Motivation

• Introduce an in-memory telemetry model to capture UI events for early MVP telemetry and offline validation.
• Wire dependency injection so view models can use telemetry and a local fake recommendations source for UI iteration.
• Replace the seeded Results flow with a recommendation card UI that surfaces trust/license/rationale and captures user actions for analytics.

Description

• Add telemetry domain and storage with TelemetryEvent, TelemetryRepository, and an InMemoryTelemetryRepository implementation and bind it via a new AppModule using Hilt.
• Enable Hilt by adding App application class, updating AndroidManifest.xml, annotating MainActivity with @AndroidEntryPoint, and converting view models to @HiltViewModel.
• Implement recommendation domain types and a FakeOerRecommendationRepository, update navigation to safely encode query (Uri.encode) and move results consumption into ResultsViewModel using SavedStateHandle.
• Replace seeded Results UI with RecommendationCard UI and actions (Use This, Need Better Match, Helpful/Not helpful) and record telemetry on view, submit, impressions, opens, ratings, and better-match requests.

Testing

• Added unit tests for InMemoryTelemetryRepository in InMemoryTelemetryRepositoryTest, which verify ordering and max-size behavior. Tests passed.
• Added HomeViewModelTest to assert HomeViewed, SearchFocused, and SearchSubmitted events and blank-submit behavior; tests passed.
• Added ResultsViewModelTest to assert ResultsViewed, impressions, SearchFailed, and action telemetry events; tests passed.
• Project compiles with the new DI and view model changes and the above unit tests were executed successfully.

---

Codex Task