XBRL: enable Doling Graph with structured graph data from Docling document

### Requested feature

XBRL (eXtensible Business Reporting Language) is an open, international standard for digital business reporting that makes financial data machine-readable.
Docling supports the parsing of XBRL instance reports from version [2.75.0](https://github.com/docling-project/docling/releases/tag/v2.75.0), through the backend parser [XBRLDocumentBackend](https://github.com/docling-project/docling/blob/main/docling/backend/xml/xbrl_backend.py#L61).

There is currently an effort to strengthen the parsing of the rich taxonomy in XBRL reports in Docling, leveraging the current implementation and keeping an eye on a refactoring of Docling's data model, which will impact the key-value pair models in [GraphData](https://github.com/docling-project/docling-core/blob/9eca661df795b9335fcf05b32705614be674e60f/docling_core/types/doc/document.py#L2470).

In order to support several uses cases for the XBRL reports (for instance, https://github.com/finos-labs/ai-evals-framework/issues/23), we would like to enable [Docling Graph](https://github.com/docling-project/docling-graph) with the pydantic models and the ability to process graph data represented in Docling document data model as structured information, so that the structure in XBRL reports can be preserved in Docling documents and reconstructed later with Docling Graph.

This issue depends on https://github.com/docling-project/docling/issues/3034.