The advantage is that you automatically get pre-loaded PyMongo or SQLAlchemy clients in your Jupyter environment, with connection URLs correctly configured to your database. You also get the same in the Streamlit app.
If you have a static dev database that you always use/maintain, and you're comfortable with your connectors, then you probably don't find a lot of value in that. But if you're like me and you always have to look up how to configure the network connections through the docker network + exposed ports, and you're never sure how to construct the connection URL and configure the default clients, then that stuff comes right out of the box.
I think these things are most useful for me in the "proof-of-concept" stage where I'm just starting a new data engineering idea from scratch and I want a brand new database.
Going along those lines - I'm going to add elasticsearch next, because I always want that search functionality over text but always feel like its hard to spin up and connect to