Flaky test: test_list_of_dataclass_update_failures::test11 fails intermittently on arm64

[huaweil-nv]

Required prerequisites

• Consult the security policy. If reporting a security vulnerability, do not report the bug using this form. Use the process described in the policy to report the issue.
• Make sure you've read the documentation. Your issue may be addressed there.
• Search the issue tracker to verify that this hasn't already been reported. +1 or comment there if it has.
• If possible, make a PR with a failing test to give us a starting point to work on!

Describe the bug

test11 in test_list_of_dataclass_update_failures (python/tests/kernel/test_assignments.py) fails intermittently on py3.11 + arm64 + cuda12.6 (Fedora 42 wheel validation). The failure is non-deterministic — it passes on re-run.
This test was previously wrapped in pytest.raises(AssertionError) because the struct argument marshaling was broken (hardcoded double type for all vector members). PR #3879 fixed the type mismatch, and the test was unwrapped. It now passes on all x86_64 configurations and most arm64 configurations, but fails intermittently on the combination above.

Steps to reproduce the bug

from dataclasses import dataclass
import cudaq

@dataclass(slots=True)
class MyTuple:
    l1: list[int]
    l2: list[int]

@cudaq.kernel
def test11(t: MyTuple, size: int) -> list[int]:
    l = [t.copy(deep=True) for _ in range(size)]
    l[0].l1 = [2]
    l[1].l2[0] = 3
    res = [0 for _ in range(4 * len(l))]
    for idx, item in enumerate(l):
        res[4 * idx] = len(item.l1)
        res[4 * idx + 1] = item.l1[0]
        res[4 * idx + 2] = len(item.l2)
        res[4 * idx + 3] = item.l2[0]
    return res

result = test11(MyTuple([1], [1]), 2)
assert result == [1, 2, 1, 1, 1, 1, 1, 3]

Expected behavior

PASS

Is this a regression? If it is, put the last known working version (or commit) here.

Not a regression

Environment

• CUDA-Q version: v0.13.0
• Python version: 3.11
• C++ compiler:
• Operating system: Fedora 42
• ** arch ** : arm64
• CUDA: 12.6
  Other arm64 combinations (e.g. py3.13 + cuda13.0) are not affected.

Suggestions

No response