README: fix H1 title + direct-install instructions for new build/ layout

Update repo-id in build.toml after migration to SNC org

Update repo-id references after migration to SNC org

README: add backward validation results, real perf numbers vs vLLM FA2

Remove obsolete torch-ext/ (replaced by build/torch-cuda/)

Restructure: build/torch-cuda/ layout with relative imports for self-contained kernels.get_kernel() use

Add README.md as the model card; remove CARD.md

Apply Dao #2484 SM120 init-time fix: pack_gqa=False on Sm120/Sm120Tma

Initial commit: 5-PR SM120 bundle validated on SM121a

initial commit