Remove full-cache scans from CP owner-lane allocation

The CP shared-KV allocator was still doing total-cache-sized CPU work in the scheduler hot path.  That cannot be hidden by GPU overlap, so owner-lane allocation now maintains per-owner free/release buckets and consumes request-sized prefixes instead of rebuilding masks over the full free-page tensor on each request.\n\nThe benchmark was extended to isolate L1 stats, selection, and allocation costs, and the CPU layout tests now install a complete sgl_kernel stub before importing SGLang helpers so remote unit collection does not abort in native extension loading.\n\nConstraint: Allocator CPU work blocks scheduler progress and cannot overlap with GPU forward execution.\nConstraint: CPU unit tests must not load native sgl_kernel on remote images where the loader can SIGABRT.\nRejected: Keep contiguous-run search over full free_pages | still scales with cache capacity and measured multi-ms overhead.\nRejected: Treat remote collection abort as an environment-only issue | it prevented allocator regression coverage and was fixable with a test-local stub.\nConfidence: high\nScope-risk: moderate\nDirective: CP owner-lane allocation is bucket-based; do not reintroduce full free_pages scans on the hot path without benchmark evidence.\nTested: Local py_compile for touched files\nTested: Local benchmark unit test, 6 passed\nTested: Remote benchmark unit test, 6 passed\nTested: Remote test_alloc_pages_with_owners.py, 10 passed\nTested: Remote test_cp_shared_kv_layout.py, 27 passed\nTested: Remote production allocator microbench shows select/alloc p50 reduced from ms-scale to sub-ms scale\nNot-tested: Full ETE traffic run after allocator bucket change

test/registered/unit/benchmark/test_cp_hicache_allocator_bench.py

@@ -1,9 +1,11 @@
 import torch
 
 from benchmark.hicache.bench_cp_hicache_allocator_overhead import (
+    StandaloneCPSharedPagedAllocator,
     StandaloneHostAllocator,
     _host_pages_from_gb,
     _make_host_free_slots,
+    _make_page_compute_owners,
     _parse_int_list,
 )
 
@@ -45,3 +47,35 @@ def test_host_random_fragmented_has_requested_size():
     )
     assert free_slots.numel() == 64 * 16
     assert torch.unique(free_slots).numel() == free_slots.numel()
+
+
+def test_standalone_l1_allocator_reports_owner_lane_stats():
+    allocator = StandaloneCPSharedPagedAllocator(
+        physical_pages=4,
+        page_size=2,
+        cp_size=4,
+        device=torch.device("cpu"),
+    )
+    owners = _make_page_compute_owners(5, cp_size=4, pattern="round_robin")
+
+    required, available, deficits = allocator.compute_owner_lane_stats(owners)
+
+    assert required == [2, 1, 1, 1]
+    assert available == [4, 4, 4, 4]
+    assert deficits == [0, 0, 0, 0]
+
+
+def test_standalone_l1_allocator_allocates_owner_matching_pages():
+    allocator = StandaloneCPSharedPagedAllocator(
+        physical_pages=8,
+        page_size=4,
+        cp_size=4,
+        device=torch.device("cpu"),
+    )
+    owners = [0, 1, 2, 3, 0]
+
+    selected = allocator.alloc_pages_with_owners(owners)
+
+    logical_pages = (selected.view(-1, allocator.page_size)[:, 0] // allocator.page_size)
+    selected_owners = torch.remainder(logical_pages - 1, allocator.cp_size).tolist()
+    assert selected_owners == owners