Make the HTML document integration structure consistent with the PDF document
integration by:

 - Extracting the integration implementation into a `HTMLIntegration`
   class in `src/annotator/integrations/html.js`
 - Renaming the document metadata extraction class from `DocumentMeta`
   to `HTMLMetadata` (to match PDFMetadata) and moving it to
   `src/annotator/integrations/html-metadata.js`