Why this !== check?

Talked it out, the other case is handled separately

So the way this works is:

1. If node is an import from another document, go resolve in that document instead. This logic already has its own handling to make sure that in the foo as bar case, we look for foo in that other document, instead of bar.

2. Otherwise, look for a feature with the identifier we're resolving in the local scope of node. This is why the export {foo} case already worked without any special handling for that syntax (we know foo and we know the scope). foo as bar didn't work because we didn't do the bar -> foo switch in this local document case.

I took a little stab at unifying the bar -> foo logic between the import and local cases, but it didn't seem to be actually more readable without maybe a larger refactoring.