Fix empty and duplicated glossary breadcrumb items

[LukasWallrich]

Why this needs addressing

The live JSON-LD on https://forrt.org/glossary/english/h_index/ currently contains:

• position 2, name: "Glossary", with an empty item;
• position 3, another name: "Glossary", pointing to /glossary/english/.

This comes from layouts/partials/structured-data/breadcrumbs.html walking the stub /glossary/ ancestor and the language section independently.

Requested change

After the glossary hub is repaired, make the breadcrumb trail explicit for glossary pages:

• FORRT → Glossary (the real hub URL) → English/German/Chinese/Arabic/Turkish → term;
• every ListItem has a non-empty, canonical item URL;
• labels are unique and describe their level;
• positions are sequential and the final item is the current page.

Apply equivalent validation to curated resources and other deep sections.

Acceptance criteria

• No generated breadcrumb contains an empty item or duplicated level label.
• Representative glossary pages pass JSON parsing and the Rich Results Test.
• The fix does not create multiple competing BreadcrumbList blocks.

Reference: https://developers.google.com/search/docs/appearance/structured-data/breadcrumb

[LukasWallrich]

Generation-pipeline clarification

This is a Hugo breadcrumb/template fix. Do not patch generated glossary term Markdown. The generated content should remain reproducible from the source sheet and _create_glossaries.py.