Model deployment

[kdaniel21]

Hi, first of all, thank you for your work, the library is awesome!

Yesterday, I opened a Q&A discussion in the main repo, wondering about what the best way would be to deploy Docling in production.

We'd love to start using Docling as soon as possible, and we'd be open to contribute as well. Hence, the question is whether you already have something in mind (or implemented), or whether you would be open to changes that'd enable deploying the models separately (e.g. on SageMaker/AzureML).