Duplicate namespec in Person.namespecs()

When listing a Person's namespecs, if there is a volume with frontmatter, 2 namespecs have the volume as their `parent` while none have the frontmatter Paper.

This happens for both old-style and new-style IDs:

```py
>>> p = anthology.get_person('shay-b-cohen')
>>> [ns for ns in p.namespecs() if ns.parent.full_id=='W15-15']
[NameSpecification(Name('Shay', 'Cohen'), id='shay-b-cohen'), NameSpecification(Name('Shay', 'Cohen'), id='shay-b-cohen')]
>>> [ns for ns in p.namespecs() if ns.parent.full_id=='W15-1500']
[]
>>> [ns for ns in p.namespecs() if ns.parent.full_id=='2024.scalellm-1']
[NameSpecification(Name('Shay', 'Cohen'), id='shay-b-cohen'), NameSpecification(Name('Shay', 'Cohen'), id='shay-b-cohen')]
>>> [ns for ns in p.namespecs() if ns.parent.full_id=='2024.scalellm-1.0']
[]
```

In fact these are identical namespecs:

```py
>>> a, b = [ns for ns in p.namespecs() if ns.parent.full_id=='W15-15']
>>> a is b
True
```

From a user perspective, this is confusing—why list the same namespec twice?

Start at the Person.namespecs() entry point and reproduce the examples for Shay B. Cohen using both old-style and new-style IDs. Trace why a volume's frontmatter produces two identical namespecs with the volume as parent instead of one associated with the frontmatter Paper. Done means each namespec is listed once with the correct parent in both cases.

由索引模型根据 Issue 内容生成。