I don't write much Go anymore but my hunch is that it's a combination of the package layout and auto method delegation for embedded structs. Even Java does a much better job of helping the developer at obviating the interfaces between different subsystems.
Python does the file level thing and it's source of constant annoyances with cyclical imports. Ugh, wasted so much hours on fixing it.
I don't love dealing with Java, but it has none of these particular issues. It's easy to find the definition of a class, because it enforces that the file path must match the fully qualified class name for all public classes (and this tends to be followed for privates as well in practice). Finding the implementation isn't quite trivial, but if you know a class in the tree then you can easily find the whole parent tree by following the "extends Foo" clauses, or find subclasses by grepping for it instead.
But that aside, Go's incessant insistence on structural typing makes it impossible for any tool to generate a list that is both complete and free from false positives.