Goose3

* Catch additional PIL exceptions when attempting to read images; see 42
* Better meta processing of opengraph tags for use as keys in returned data; see 45

* Changed configuration to ***not*** pull images by default [31](https://github.com/goose3/goose3/issues/31)
* Update `get_encodings_from_content` to return a string and remove trailing spaces [35](https://github.com/goose3/goose3/pull/35)
* Remove infinite recursion on parser selection [39](https://github.com/goose3/goose3/pull/39)
* Document video and image classes
* Re-add remaining image tests

* Add `soup` as a parser option to use `lxml.html.soupparser` [27](https://github.com/goose3/goose3/issues/27)
* Fix an issue with passing the requests session object to the crawler
* Documentation on readthedocs (https://readthedocs.org/projects/goose3/)
* Code cleanup

* Fix using different `requests` session for each url fetched
* Added `close` method to the Goose object
* Allow the Goose object to be a context manager
python
from goose3 import Goose
with Goose() as g:
g.extract(url='some-url-here')

***NOTE:*** No need to change code as it will attempt to automatically close
the connection on garbage collection
* Configuration object changes
* Better handling of the `known_context_patterns` configuration
* Added http_headers configuration option to be passed to `requests`
* Added http_proxies configuration option to be passed to `requests`
* Added http_auth configuration option to be passed to `requests`
* Fix base64 image parsing 7

* Fix installation issue
* Removed unused/broken regex
* Include all necessary files
* Fix failed tests (**most**)
* Resolved relative URL issue 21
* Resolved temporary files not being properly removed 18
* Removed unused dependencies and code to support python 2 16
* Fix error when using the configuration object to configure goose 14