Provide auto-complete in notebooks for table names starting with an underscore

[nanne-aben]

Suppose we'd make this table.

df.write.saveAsTable("default._person")

And then we try to load it using Databases() (or any of the related classes).

db = Databases()

Then we don't get auto-complete, because our notebook assumes that we don't want to access private variables. To circumvent this, we rename the attribute as such:

person, _person = db.default.u_person()

The underlying table is not renamed, solely the class attribute used for autocomplete.

When renaming the attribute leads to a naming conflict (e.g. because u_person already exists), we resolve the conflict by adding more underscores (e.g. _person would then become u__person).