ycb_django_pr13158

what a new run launched now would send

QuerySet.none() on combined queries returns all results.
Description
	
I came across this issue on Stack Overflow. I'm not 100% sure it's a bug, but it does seem strange. With this code (excuse the bizarre example filtering):
class Publication(models.Model):
	pass
class Article(models.Model):
	publications = models.ManyToManyField(to=Publication, blank=True, null=True)
class ArticleForm(forms.ModelForm):
	publications = forms.ModelMultipleChoiceField(
		Publication.objects.filter(id__lt=2) | Publication.objects.filter(id__gt=5),
		required=False,
	)
	class Meta:
		model = Article
		fields = ["publications"]
class ArticleAdmin(admin.ModelAdmin):
	form = ArticleForm
This works well. However, changing the ModelMultipleChoiceField queryset to use union() breaks things.
publications = forms.ModelMultipleChoiceField(
	Publication.objects.filter(id__lt=2).union(
		Publication.objects.filter(id__gt=5)
	),
	required=False,
)
The form correctly shows only the matching objects. However, if you submit this form while empty (i.e. you didn't select any publications), ALL objects matching the queryset will be added. Using the OR query, NO objects are added, as I'd expect.

Work only inside this repository checkout. Make the code change the task
describes, keeping the diff focused — no drive-by refactors.

When you are done, leave your changes committed or in the working tree;
they are collected automatically.

Stay on this snapshot checkout (`task/ycb_django_pr13158`). Never checkout, pull, or rebase onto `main`. That branch is a README-only orphan.
Stay on this HEAD. Do not fetch another default branch. Push only on the Cursor-created `crazy-cursor/…` side branch from this HEAD.

Some past runs of this task were launched with a different prompt (the prompt template changed since, or those runs predate this benchmark's stored prompt). Each run persists the exact prompt it sent at launch — that per-launch record is the audit trail; this page shows only the current one.