feat: Add DoclingDocument API in new docling-core module

[ThomasVitale]

Fixes gh-98

---

I tried to map the structure of the Docling project, which uses a dedicated Docling Core package for data types models: https://github.com/docling-project/docling-core. So I introduced a new docling-core module, including some basic documentation. Since we are using the ai.docling.api.serve and ai.docling.client.serve packages for the other two modules, I went with ai.docling.api.core for consistency. But I'm open to other solutions (e.g. ai.docling.core, ai.docling.serve.api and ai.docling.serve.client).

I have added an explicit DoclingDocument class following the same approach used for the Serve APIs: Lombok + Jackson 2 and 3 compatibility. This is probably gonna be one of the hardest classes to maintain manually, but until we come up with a reliable automated process, I guess we'll have to. I initially thought about making it a bit more generic (e.g. using simple Strings rather than enums), but that would reduce its usefulness by a lot. In order to make it useful for downstream scenarios (e.g. integrations with LangChain4j or Spring AI for RAG and agent workflows), we need to have a full type-safe representation.

Finally, I considered how to switch from Map to DoclingDocument in a backward compatible way in the context of DocumentResponse, but in the end I thought it wasn't worth the effort considering we're at very beginning and using the returned Map is pointless without a proper data type. So, I switched DocumentResponse.getJsonContent() to return a DoclingDocument instead of Map<String, Object>.

Thoughts? @edeandrea @lordofthejars

[github-actions]

	Tests	Passed ✅	Skipped	Failed
Gradle Test Results (all modules & JDKs)	336 ran	336 passed	0 skipped	0 failed

Test	Result
No test annotations available

• View Gradle Test Results (all modules & JDKs)

[github-actions]

HTML test reports are available as workflow artifacts (zipped HTML).

• Download: Artifacts for this run

[edeandrea]

Thanks @ThomasVitale!

Finally, I considered how to switch from Map to DoclingDocument in a backward compatible way in the context of DocumentResponse, but in the end I thought it wasn't worth the effort considering we're at very beginning and using the returned Map is pointless without a proper data type. So, I switched DocumentResponse.getJsonContent() to return a DoclingDocument instead of Map<String, Object>.

I agree. I think we can still make breaking changes for a bit.

I'm also going to be OOO the rest of this week, so don't let me be a bottleneck if you are planning to work on this. I trust your judgement - feel free to merge things.

[github-actions]

HTML test reports are available as workflow artifacts (zipped HTML).

• Download: Artifacts for this run

[ThomasVitale]

Thanks for the review, @edeandrea!

[github-actions]

HTML test reports are available as workflow artifacts (zipped HTML).

• Download: Artifacts for this run