feat: unify attention backend with multi-dim mask support

- Add attention() functional entry delegating to active backend
- GQA/MLA forward calls attention() instead of inline cache/SDPA
- CUDA kernels support 2D/3D/4D mask via mask_h_stride field
- CudaBackend.fwd_decode builds 2D padding mask for mixed seq_lens
- KVCache.max_len precomputed in bind_tasks to avoid GPU sync
- batch==1 decode short-circuits mask=None
- Split tests into conftest, test_backend, test_backend_equivalence, test_kernel_mask
- 440 tests pass, L20 decode 1.44-1.60x speedup vs torch native

csrc/kernels/attn_paged_decode_split_kv.cuh

@@ -31,7 +31,7 @@ __global__ void paged_attn_decode_split_kv_kernel(PagedAttentionParams<bf16> p)
     int ch_begin = split * chunks_per_split;
     int ch_end = min(chunks_total, ch_begin + chunks_per_split);
 
-    const int mask_base = batch * p.mask_b_stride;
+    const int mask_base = batch * p.mask_b_stride + q_head * p.mask_h_stride;
 
     for (int ci = ch_begin; ci < ch_end; ci++) {
         int chunk_start = ci * PDC_CHUNK;