Add CUDA 13.4 Rubin SM 10.7 support

[scal444]

No description provided.

[greptile-apps]

Confidence Score: 5/5

The PR appears safe to merge, with no concrete correctness, security, build, or repository-rule violations identified.

Summary

• Adds SM 107 to CMake architecture selection and capability definitions.
• Enables SM 107 tensor-operation dispatch for similarity kernels.
• Adds Rubin-specific shared-memory and thread-capacity constants.
• Constrains GSI static shared-memory sizing for SM 107.

_{Reviews (1) · Last reviewed commit: "Add CUDA 13.4 Rubin SM 10.7 support"}