Prepare ColocBoost 1.0.10: scalability, robustness, and tutorials

[xueweic]

Summary

This PR prepares ColocBoost 1.0.10 with improvements to computational scalability, robustness, and conceptual documentation. It also addresses the CRAN-reported MKL-sensitive test failure in version 1.0.9.

No package dependencies are added or changed.

Computational scalability

• Reduced redundant LD calculations, purity evaluations, matrix operations, and memory reallocations.
• Improved computational efficiency for analyses involving large numbers of phenotypes.

Robustness fixes

• Improved handling of outcome names and large trait configurations.
• Addressed edge cases in CoS assembly, output generation, and plotting.
• Stabilized the CRAN-reported MKL-sensitive tests using stronger simulation signals and aligned test inputs, without skipping the affected tests.

Documentation and tutorials

• Added a conceptual framework for multi-trait colocalization events and interpretation of CoS, VCP, and NPC.
• Added advanced examples covering multiple causal variants and weaker disease-GWAS signals.
• Added six high-level summary figures with empirical results and reorganized vignette-only assets to reduce duplicated installed-package content.