refactor: extract shared dispatcher header, unify MMA/scalar dispatch format

- Merge 3 duplicated dispatch blocks into single attn_dispatchers.cuh
- Merge compute_num_splits from attn_utils.cuh into dispatcher header
- All dim3 grid/block declarations and <<<>>> launches are single-line
- Production .cu files (35-42 loc) only handle torch wrapping + pybind11
- Test files include dispatcher header directly, removing all #ifndef ASTRAI_NO_MMA duplication

csrc/kernels/attn_entry_utils.cuh

@@ -5,13 +5,6 @@
 
 using bf16 = __nv_bfloat16;
 
-inline int compute_num_splits(int base_blocks, int tiles_total) {
-    int sm_count = 0;
-    cudaDeviceGetAttribute(&sm_count, cudaDevAttrMultiProcessorCount, 0);
-    int n = (2 * sm_count + base_blocks - 1) / base_blocks;
-    return std::max(1, std::min(n, std::min(tiles_total, 32)));
-}
-
 // Dispatch head_dim: shared macro — avoids C++20 lambda template syntax.
 // Usage: DISPATCH_HEAD_DIM(hd, fn, arg)
 //   Expands to: fn<32>(arg); fn<64>(arg); etc.