perf: extend MMA decode to arbitrary GQA ratio, add launch bounds, vectorize combine

- Multi-pass MMA: encode pass in grid blockIdx.x, compute q_head0/G in-kernel
- Fixes crash for G>32 (previously block(32,G) exceeded 1024 threads)
- Fixes alloc_split_partials using uninitialized num_splits (MAX_SPLITS=32)
- __launch_bounds__ on all MMA and prefill kernels for better register allocation
- 4x vectorized combine kernel (4 head_dim per thread)
- uint4 vectorized K loads in scalar decode kernels
- cp.async .L2::128B cache hint for K/V tile streaming
- Extract warp_reduce_sum, bf16, MAX_SPLITS to attn_warp_utils.cuh

csrc/kernels/attn_entry_utils.cuh

@@ -2,6 +2,7 @@
 #include <torch/extension.h>
 #include <c10/cuda/CUDAGuard.h>
 #include "attn_common.h"
+#include "attn_warp_utils.cuh"
 
 using bf16 = __nv_bfloat16;
 
@@ -22,8 +23,8 @@ using bf16 = __nv_bfloat16;
 template<typename P>
 inline void alloc_split_partials(P& p) {
     auto fopt = torch::TensorOptions().dtype(torch::kFloat32).device(torch::kCUDA);
-    auto o_part = torch::empty({p.batch, p.q_head, p.num_splits, p.head_dim}, fopt);
-    auto ml_part = torch::empty({p.batch, p.q_head, p.num_splits, 2}, fopt);
+    auto o_part = torch::empty({p.batch, p.q_head, MAX_SPLITS, p.head_dim}, fopt);
+    auto ml_part = torch::empty({p.batch, p.q_head, MAX_SPLITS, 2}, fopt);
     p.o_part = (float*)o_part.data_ptr();
     p.ml_part = (float*)ml_part.data_ptr();
 }