Migrate clinical-data-enrichments endpoint to clean architecture with ClickHouse support

[fuzhaoyuan]

Built on top of #11425

Refactor Clinical Data Enrichments

Summary

Refactor clinical data enrichments to follow clean architecture principles, migrating from legacy service layer to domain layer with ClickHouse performance optimizations.

Note

In the event of responses largely differ from legacy responses, it's due to the legacy implementation doesn't unify casing of attribute values, causing inaccurate number of groups, for example:

This is because legacy has both 'PRIMARY_SITE' attribute values of "Ascending Colon" and "Ascending colon" as separate groups, having one more group than needed, thus led to inaccurate calculation results.

Key Changes

Architecture Migration

• Before: Logic in org.cbioportal.legacy.web.ClinicalDataEnrichmentController + ClinicalDataEnrichmentUtil
• After: Clean separation across layers
  • REST: ColumnStoreClinicalDataEnrichmentController
  • Domain: FetchClinicalDataEnrichmentsUseCase (business logic)
  • Domain: ClinicalDataEnrichmentUtil (pure statistical functions)

Performance Improvements

Legacy approach - Multiple queries per group:

// For each group, separate DB calls
groupedSamples.stream()
    .map(group -> clinicalDataService.fetchClinicalData(...))

New approach - Single optimized query:

// Fetch ALL data for ALL groups in ONE query, regroup in-memory
fetchClinicalDataSummaryForEnrichments(allSampleIds, allPatientIds, ...)

Impact: Reduces database queries from O(groups × attributes) to O(1)

New Domain Layer

• ClinicalDataEnrichment - immutable record (was mutable bean)
• EnrichmentTestMethod - type-safe enum (was magic strings)
• FetchClinicalDataEnrichmentsUseCase - encapsulates enrichment workflow
• ClinicalDataEnrichmentUtil - pure statistical functions

Repository Enhancements

• fetchClinicalDataSummaryForEnrichments() - batch fetch with UNION query
• getClinicalDataCountsForEnrichments() - optimized categorical counts
• New ClickHouse SQL mapper handling sample/patient/conflicting attributes

Other Improvements

• Comprehensive test coverage
• MapStruct DTO mapping
• Fixed PreAuthorize for ClinicalData and ClinicalDataEnrichment
• Better integration test organization