2.16

Added

• Many, many comics :D
• Comic modules can now use a proper HTML parser (based on lxml) with XPath
  or CSS expressions.
• Users are notified if they are using modules which were
  removed/moved/renamed.
• Use CodeClimate to analyse source code quality and CodeCov to track test
  coverage.

Changed

• Annotation text can now be optional (Fetching a comic doesn't fail if it
  doesn't appear on every page).
• When using HTML output, next and previous links are now also at the end of
  the page.
• Use the Python import system (previously PEP-302) instead of the file system
  to find modules.
• Remove embedded colorama, depend on it instead.
• Replace homegrown PY2/3 compatibility with six.
• Build dosage website with staticsite.
• Reworked comic module structure. Webcomics are now class instances instead of
  classes. While this doesn't change that much for single comic modules,
  "virtual" modules covering multiple comics can now be written much more
  concise. See issue #42 for details.
• Windows build is now one standalone EXE file (thanks to PyInstaller).
• Allow combining -n with -c or -a (related to #90). This allows to set an
  upper bound to normally "unconstrained" fetch modes.
• Replace forced "sleep" between requests with a simple host-based throttling
  mechanism.
• Uses imagesize instead of Pillow to get image sizes. (Pillow is a very big
  libary with many binary dependencies and we were using only this feature from
  it)

Removed

• Many, many broken/disappered comics :(

Fixed

• Many, many comics :D
• Let Requests figure out encodings.
• Don't crash when HTML output is run more than once per day. Closes: wummel#78