DA3METRIC-LARGE parallel reasoning and confience missing

Hello author, thank you very much for such a great open source project ！！！
I have two questions when using the model DA3METRIC-LARGE
The first issue is that I want to integrate DA3 into reinforcement learning, where each agent inputs one RGB image (multiple uncorrelated images for parallel inference, input in batch size). However, I found that the model does not seem to provide an interface for this parallel inference
The second issue is that I did not have confidence output when using the DA3METRIC-LARGE model. I want to use confidence to filter out the depth blur caused by solid color backgrounds

The issue names the DA3METRIC-LARGE model but no files, tests, or entry points. Start by locating its inference interface and output structure, then determine how batched uncorrelated RGB inputs and confidence values should be exposed. Done means parallel inference works and confidence can be used to filter depth blur from solid-color backgrounds.

Written by the indexing model from the issue text.