v1.0.2: Article bylines, scraper fallbacks fixed

Added

• Article bylines (articles.author). Taken from the feed or NewsAPI
  entry, or from the publisher's page when the scraper reads it, and cleaned
  ("By Jane Doe, CNN" becomes "Jane Doe"; "Staff" and outlet names are left
  blank). Not shown on pages yet. Upgrading adds one empty column
  (./install.sh applies the migration); articles stored before it have no
  byline.

Fixed

• The scraper's page-reading fallbacks run again. Since 0.5 the first
  page it downloaded was thrown away, so three fallbacks never ran: finding
  the AMP version of a page from its link, reading the article from the
  page's structured metadata, and (new here) the byline from the page.
  Sites without a usable feed author or a clean readable page get more of
  each now.