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

astrai/extension/__init__.py

@@ -20,6 +20,7 @@ from astrai.extension.attention_backend import (
     AttentionBackend,
     CudaBackend,
     TorchNativeBackend,
+    attention,
     attn_backend,
     get_backend,
 )
@@ -35,6 +36,7 @@ __all__ = [
     "AttentionBackend",
     "CudaBackend",
     "TorchNativeBackend",
+    "attention",
     "attn_backend",
     "get_backend",
     "attn_decode",