refactor: reorganize CUDA kernels into per-family directories

- move attention kernels to csrc/kernels/attention/ and rotary to rotary/
- add shared common/mma.cuh (mma_sync, ldmatrix) and device.cuh (sm checks)
- split fp8_mm into three-layer fp8/common.h, gemm.cuh, mm.cu
- fix fused FP8 GEMM ldmatrix lane indexing to fix OOB shared reads
- update extension ops, loader, and kernel tests

csrc/kernels/attention/warp_utils.cuh

@@ -0,0 +1,13 @@
+#pragma once
+#include <cuda_bf16.h>
+
+using bf16 = __nv_bfloat16;
+
+static constexpr int MAX_SPLITS = 32;
+
+__device__ inline float warp_reduce_sum(float val) {
+    #pragma unroll
+    for (int offset = 16; offset > 0; offset >>= 1)
+        val += __shfl_xor_sync(0xFFFFFFFF, val, offset);
+    return val;
+}