Enhance Similarity Maps (Embeddings) Visualization

cBioPortal introduced Similarity Maps, a feature that visualizes UMAP/PCA embeddings of cancer samples, allowing researchers to discover hidden patterns across thousands of samples simultaneously. However, the current implementation is a prototype that fetches data directly from a hardcoded S3 URL, bypassing cBioPortal's data pipeline entirely and following no established datahub standards. This project makes Similarity Maps production-ready by building a proper data layer that follows cBioPortal's existing architecture. This includes designing a datahub file format for embeddings, reusing/extending the existing importer to load embedding data into ClickHouse, and exposing REST API endpoints that replace the hardcoded S3 fetch. The frontend will then be updated to consume the new API dynamically. Additionally, enhanced tooltips and improved colouring options will be implemented to strengthen the visualization. Key deliverables include: an approved database schema, a working EmbeddingImporter.java, a validated datahub file format, REST API endpoints, updated frontend integration, and full documentation