Private extractors

[lmarini]

Ability to deploy extractors that only a specific users can run. Currently any extractor that is deployed to an instance will be accessible to every user on the system. This means that administrators always have to be incredibly careful about deploying third party extractors to a specific instance.

This would allows us to let users provide code at runtime that they want to run on their data. For example, they could write a python script in a jupyter notebook and we could deploy that as an extractor to run on demand on data.

Another use case would be to let us pull in preprocessing for hugging face at runtime and let the user deploy a model that requires preprocessing.

[ddey2]

Any though on how we can make it private?

I was thinking of having the list of users or list of groups as a field in the listener collection in the db and adding the same in extractor_info.json. Lmk your thoughts.