add <description> and <summary>

[vagenas]

• Added optional <description> and <summary> elements to the element head in the spec documentation.
• Updated relevant sections in the XSD and Schematron to enforce the correct order of these elements.
• Update test suite accordingly
• Document diffences between <caption>, <description>, and <summary>
• Fix minor typo in recommended token vocabulary

[vagenas]

Mental model:

	<caption>	<description>	<summary>
Provenance	from the original document	derived (VLM, human, pipeline)	derived
Function	labels the component	accounts for the component	distills the component
Question answered	“What does the document call this?”	“What is this / what does it show?”	“What is the gist / takeaway?”
Typical source	OCR / layout extraction	VLM, annotator	summarizer, indexer, RAG prep
Content	raw text + semantic elements	raw text only	raw text only

Length can overlap across all three — the split is purpose, not brevity.

One-line definitions:

• <caption> — document-emitted label associated with the component
• <description> — derived literal account of the component’s content
• <summary> — derived abstractive condensation of what the component conveys

[PeterStaar-IBM]

lgtm!

[dolfim-ibm]

lgtm