feat: add comprehensive real-world test datasets with VCF and BAM files

[berntpopp]

Summary

Add realistic test datasets to replace toy data with production-scale VCF and BAM files for comprehensive testing.

Current Limitations

• Small toy VCF files (~10-50 variants)
• No BAM file integration for IGV reports
• Missing complex inheritance patterns
• Limited annotation coverage

Proposed Datasets

1. Clinical Trio - 45K variants, rich annotations, BAM files
2. Research Cohort - 50 samples, 2M variants, population genetics
3. Cancer Analysis - tumor-normal pairs, somatic variants
4. Performance Benchmark - 1M+ variants for scalability testing

Implementation

• Cloud storage integration (S3/Azure)
• On-demand download for CI/CD
• Realistic file sizes (100MB-10GB)
• Comprehensive annotations (gnomAD, CADD, ClinVar)

Tasks

• Set up cloud storage for large datasets
• Create clinical trio dataset with BAM files
• Generate research cohort for population genetics
• Add performance benchmark datasets
• Implement dataset manager with lazy loading

Success Criteria

• 5 realistic datasets covering major use cases
• BAM integration for IGV report testing
• Performance benchmarks with production-scale data

Labels: enhancement, testing, data, infrastructure