Multiple GPU support

Is there a way to run DoclingLoader or a pipeline on multiple GPUs simultaneously?

Another question:
Is it possible to run DoclingLoader with GPU support in different threads? I need to process multiple documents concurrently within the same stack.

Start by reading the DoclingLoader and pipeline entry points to understand how GPU configuration and document processing are currently handled. Reproduce concurrent processing with multiple documents, then define and verify expected behavior for multiple GPUs and separate threads; done means the supported behavior is implemented and documented with coverage for these cases.

Written by the indexing model from the issue text.