Mf2py

- Implied property checks now ignore alt="", treating it the same as
if no alt value is defined.
- Support for using a custom dict implementation by setting
mf2py.Parser.dict_class. collections.OrderedDict yields much nicer
output for hosted parsers.

- Performance improvement changing simple calls to soup.find_all to
a manual iteration over .contents.

- Performance improvement by limiting number of calls to soup.find_all
in backcompat module. Should not be any functional changes.

- Backward compatibility parsing for rel=tag properties. These are now converted
to p-category based on the last path segment of the tag URI as spec'd in
http://microformats.org/wiki/h-entry#Parser_Compatibility
- Optional property html_parser to specify the html parser that BeautifulSoup
should use (e.g., "lxml" or "html5lib")

- `u-*` properties are now parsed from `<link>` elements per the updated spec
http://microformats.org/wiki/microformats2-parsing-issues#link_elements_and_u-_parsing

- Version number bumped to 1.0.0 following community discussion.