British address parser + test case fixes

This is a bit of a bumper merge request, but it would be great to get this merged and released. The highlights include:

A new British address parser
Increase number of valid street types in the US parser to all those accepted by USPS (see https://pe.usps.com/text/pub28/28apc_002.htm)
Updated the text normalisation to substitute newlines for commas, this allows more of the structure of the address to be seen in British addresses as these are often written over multiple lines.
Fix bug in tests where failing tests actually passed because the value was not checked (eg post_direction would match 'N.', but not the full 'N. ')
Fix a typo in tests where street_name test cases were actually tested against the street_number regex
Added more tests, test cases and hopefully made test code more robust with a single function to execute the tests

There are some more details in the commit messages.

Please let me know if you have any feedback and thanks for the great package!

vladimarius / pyap