Metadata-parser

* black via pre-commit
* upgraded black; 20.8b1
* integrated with pre-commit
* github actions and tox
* several test files were not in git!

updated docs on bad data
black formatting
added pyproject.toml
moved BeautifulSoup generation into it's own method, so anyone can subclass to customize
:fixes: https://github.com/jvanasco/metadata_parser/issues/25
some internal variable changes thanks to flake8

added some docs on encoding

clarifying some inline docs
BREAKING CHANGE: `fetch_url` now returns a tuple of `(html, encoding)
now tracking in ParsedResult: encoding
ParsedResult.metadata['_internal']['encoding'] = resp.encoding.lower() if resp.encoding else None
`.parse` now accepts `html_encoding`
refactored url fetching to use context managers
refactored url fetching to only insert our hooks when needed
adjusted test harness to close socket connections

better Python3 support by using the six library

added tests for url entities
better grabbing of the charset
better grabbing of some edge cases