Storing varied data in one index

What is the correct way to store varied data in an index?
I want to store multiple documents, and other texts in the same index (since I using the free plan for testing).
But it seems like this is very ineffecient.
I’d prefer a way to limit searchs to a subset of documents.
I can add a source, title or subject metadata (or all of them), but these will have unique values per document. I understand that will make search slow.
Also, adding the document itself seems even more wasteful, as it will take more memory and slow everything futher because it will be indexed.

Adding user support on that (separate data per user) seems even more complicated.

I feel I’m missing something, but couldn’t find instructions for anything other than the most simple case in the documentation.

Have you looked at using name spaces?

You can make a name space for each different document you want to index and then query specific name spaces.

I did, then I’ve found explanations here that a namespace is not more effective than filtering by metadata. So what’s the point?

And even if i do so, how to split them?
Name space per user? Per subject? A combination?