refactor: remove bf16 gemm and swiglu kernels and rebuild csrc benchmarks

- delete csrc/kernels/gemm.cu and swiglu.cu and drop their CMake and setup.py registration
- remove the ops wrappers plus backend/linear.py and backend/swiglu.py so Linear and MLP call F.linear directly
- drop the four gemm and swiglu kernel test files and prune the stale cuda_kernels.md sections
- add csrc/bench benchmarks for the remaining kernels: attention decode prefill paged decode paged prefill versus single-launch SDPA references, rotary versus the torch fallback, fp8 quantize and mm_fp8 versus torch baselines
- attention, rotary_emb, and fp8_ops kernels are unchanged

setup.py

@@ -121,8 +121,6 @@ class _CMakeBuildExt(_build_ext):
             "attn_prefill",
             "attn_paged_decode",
             "attn_paged_prefill",
-            "bf16_gemm",
-            "bf16_swiglu",
             "rotary_emb",
         )
         missing = [name for name in required if not any(lib_dir.glob(f"{name}.*.so"))]