[Feat]: for the data model of a2a, adopt the semantics of existing standards (ADMS, DCAT etc)

### Is your feature request related to a problem? Please describe.

In the context of LLMs and AI, knowledge graphs have gotten renewed interest. There is a huge body of standards describing metadata in existing (rdf based) knowledge graphs. In order for this to become the 15th standard, it would be good to allign the data model with w3c recommendations that describe most of the information requirements used here (at the very least for the agent cards). This would clarify a lot of the semantics of the data model and improve interoperability with existing applications

### Describe the solution you'd like

Ideally the a2a data model should adopt as much as the semantics described in:

- ADMS for models and agents: https://www.w3.org/TR/vocab-adms/,
- DCAT for datasets and data services: https://www.w3.org/TR/vocab-dcat-3/
- prov-o for any audit/lineage metadata: https://www.w3.org/TR/prov-o/
- dublin core terms for common/generic metadata attributes: https://www.dublincore.org/specifications/dublin-core/dcmi-terms/

Looking at the current specification, there is a lot of commonality between A2A and the above standards (which are designed to complement each other)

### Describe alternatives you've considered

While I adoption of existing standardized semantics should the end goal (for 15th standard reasons) a good intermediate option would be to provide a mapping (preferably through a json-ld context https://www.w3.org/TR/json-ld11/) to improve interoperability

### Additional context

especially dcat is a commonly used standard for data catalogs, Amongst others, the EU's open data portal (data.europa.eu) as well as many national government data catalogs are based on this standard.

### Code of Conduct

- [x] I agree to follow this project's Code of Conduct