feat: SGLang-style paged attention kernels replace page-table path

- PagedAttentionParams uses flat KV pool + req_to_token + kv_indptr/qo_indptr instead of page_table
- MMA split-KV decode and split-Q prefill kernels with indirect ragged-batch addressing
- Prefill kernel accepts 4D mask (causal-aware); decode kernel supports 2D mask
- CudaBackend is inference-only: kv_cache=None raises, no torch fallback
- benchmark.py: required --ckpt, --backend/--compare options
- Parallel build isolates build-temp/build-lib per subprocess
- Standalone test covers decode/prefill with mask, 27 cases pass

astrai/extension/loader.py

@@ -11,7 +11,13 @@ import logging
 
 logger = logging.getLogger(__name__)
 
-KERNEL_NAMES = ["attn_decode", "attn_prefill", "attn_paged_decode", "rotary_emb"]
+KERNEL_NAMES = [
+    "attn_decode",
+    "attn_prefill",
+    "attn_paged_decode",
+    "attn_paged_prefill",
+    "rotary_emb",
+]
 
 _available: dict[str, bool] = {}
 _modules: dict[str, object] = {}