test(cuda.core): wait for the memset before growing the VMM buffer

The slow-path regression test fills the buffer with cuMemsetD8 and grows
it at once. Memset is asynchronous with respect to the host, and on WDDM
the batched kernel can still be pending when the grow unmaps the old
range, so it faults with a sticky CUDA_ERROR_ILLEGAL_ADDRESS that took
every later test in the job down with it. Synchronize before the grow.

Issue #2877

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

Author: Andy-Jost

cuda_core/tests/test_memory.py

@@ -1524,6 +1524,9 @@ def test_vmm_allocator_grow_allocation_slow_path_closes_old_buffer(init_cuda):
     buf = vmm_mr.allocate(2 * 1024 * 1024)
     old_ptr, old_size = int(buf.handle), buf.size
     handle_return(driver.cuMemsetD8(old_ptr, 7, old_size))
+    # The memset is asynchronous and the grow unmaps the old range, so let it
+    # finish first; on WDDM the batched kernel otherwise faults after the unmap.
+    device.sync()
 
     # Occupy the address range right after buf so the adjacent reservation cannot
     # be honored and modify_allocation has to take the slow path.