My vision of how this might work is we have an automated tool that
visits a list of given pages on various browsers, take screenshots of
how they look and then compares the images with the last known state.
The tool checks how similar the images are and complains if they are
not the same