fix(import): check better for duplicate URIs

See #800 (closed).

I have also modified the importManuscripts mutation to not wait for the imports to complete before returning, as this was sometimes causing a timeout. Instead, it returns immediately, while the import continues in the background. That mutation used to return a list of new manuscripts, but the client didn't actually do anything with that list. So now it simply returns a boolean: true if it has started an import; false if it hasn't (either because an import is currently in progress or because there are no sources to import from). But we're not actually using that return value either, at the moment.