It would appear that Google explicitely blocks requests with a Browser
user agent if they originate from a server IP address (Google knows
about). To circumvent this we would either need to use a proxy or a
different user agent. The curl user agent seems to be working for now so
let's try and use that.

Additionally it seems that Google returns slightly altered content
depending on the originating IP address of the request. This might make
reliably unit tests impractical. If it turns out that Google takes other
measures to prevent scraping I will remove the scraper because I'm not
going to play chicken with Google.