cuSPARSE SDDMM

[yhmtsai]

Summary:
This gives the implementation for cuSPARSE SDDMM

Details:

• SDDMM is alpha(AB . spy(C)), so the output size is already known.
• Maybe I miss the SDDMM test on cpu side. Do we have it?
• additionally, I add DnMat handle to create the Dense matrix. Do we support row and col major from mdspan?

Merge Checklist:

• Passing CI
• Update documentation or README.md
• Additional Test/example added (if applicable) and passing
• At least one reviewer approval
• (optional) Clang sanitizer scan run and triaged
• Clang formatter applied (verified as part of passing CI)