Xword-dl

Quick maintenance release fixes AmuseLabs downloading, closing 81. This version uses a new mechanism for finding the key used to unscramble puzzle data, which should hopefully be more robust. Thanks to boisvert42 for proposing a fix, though if it breaks all blame belongs with me.

The marquee additions in this release: new downloaders for the McKinsey Crossword and Der Standard, contributed by iNtEgraIR2021 ­— thank you!

Beyond the headlines:
- a refactor to the download-by-URL system (described in 73) means downloaded puzzles are able to pick better filenames for themselves and select more granular settings. Mostly behind the scenes, but it immediately closes 74.
- bug fix on Universal and USA Today downloaders
- consolidation of NYT and NYT Variety parsing, now using the same API and a much smaller parser for easier maintenance
- bug fix on Guardian puzzles, now more accurately labeling when solutions are unavailable (closing 72)
- switch select-by-URL page parsing to `lxml` (a new requirement) to give a more lenient read to malformed pages. This one actually affects at least one page in the wild!

Quick bugfix release for a possible problem with pip installation.

A very cryptic update to xword-dl:

* Support added for The Globe And Mail cryptic (closing 52)
* Support added for The Guardian's various cryptics (closing 67)
* Support added for New York Times Variety puzzles (closing 59)
* Support added for embedded Crossword Compiler widgets
* Re-written README documenting filename templating, stored config file settings and more

Special thanks this release to iNtEgraIR2021 for a bugfix in 63 and to mixographer for testing and feedback on the Guardian.

New release with a major refactor and a new package organization. Notably, this new setup breaks the individual site scrapers (`downloaders`) out of one big script and into their own files in a new subpackage. My hope is that this will be easier for me to maintain, easier for outside contributors to model new scrapers off existing ones, and also easier for people who want to add and maintain scrapers that I for whatever reason deem to be outside of the scope of the main release branch.

Other bug fixes:
* Improved title handling for Friday (themed) New Yorker puzzles
* Fixed Newsday search-by-date (it needed to include a picker token, which its parent class was already capable of)
* Improves error handling throughout, and notably on New York Times puzzles where authentication is not current
* Actually includes a license! I'd always considered this free software but nice to make it official.