[Simulation.Core] WorkerThread: fix use-after-free crash with ultra-short Task

[fredroy]

Got this bug while working with a parallelization of a component, with Claude.

After investigation (not so easy even with Claude 😅), it appears that in WorkerThread,

sofa/Sofa/framework/Simulation/Core/src/sofa/simulation/task/WorkerThread.cpp

Lines 198 to 221 in 12e74ee

	bool WorkerThread::pushTask(Task *task)
	{
	// if we're single threaded return false
	if (m_taskScheduler->getThreadCount() < 2)
	{
	return false;
	}
	{
	simulation::ScopedLock lock(m_taskMutex);
	const int taskId = task->getStatus()->setBusy(true);
	task->m_id = taskId;
	m_tasks.push_back(task);
	}
	if (m_taskScheduler->testMainTaskStatus(nullptr))
	{
	m_taskScheduler->setMainTaskStatus(task->getStatus());
	m_taskScheduler->wakeUpWorkers();
	}
	return true;
	}

if a task is so fast that it finished before the line 216, task has been deleted and is garbage. So it crashes (or some random stuff)

To demonstrate the bug, I asked Claude to write a "unit test" which crashes, at least on my computer (macbook pro m3).
I am wondering if the CI (and its slow multi-threading config, if any) will trigger the bug. 🤔
I will push the fix shortly after in this PR, because we cannot merge a crash 😅

EDIT: It does crash on the macOS CI, most certainly because the CI is bare metal and much faster.

[with-all-tests]

---

By submitting this pull request, I acknowledge that
I have read, understand, and agree SOFA Developer Certificate of Origin (DCO).

---

Reviewers will merge this pull-request only if

• it builds with SUCCESS for all platforms on the CI.
• it does not generate new warnings.
• it does not generate new unit test failures.
• it does not generate new scene test failures.
• it does not break API compatibility.
• it is more than 1 week old (or has fast-merge label).