Hi! We're using ClearML in our pipeline and have a question regarding datasets.
Currently, we export training data directly from ClickHouse to S3 using a command like:

INSERT INTO FUNCTION s3(...) SELECT ...

So the data goes straight to S3 without being saved locally.
Is it possible to register this already-uploaded file in S3 as a ClearML Dataset without re-uploading it (i.e., just referencing the existing S3 file)?
If so, what's the recommended way or API to do that?
Thanks!