calc_expected_status_length does not work #415

calc_expected_status_length is broken in two ways right now.

1. URL_REGEXP only recognizes URLs at the start of a string, which is correct for is_url, but for calc_expected_status_length, all URLs should be detected, not just URLs at the start of the tweet. There should be a different URL_REGEXP for calc_expected_status_length without the start-of-string makers.
2. The URL regex has multiple groups, so findall returns a list of tuples, not strings. If there are matches, replaced_chars = len(''.join(match)) crashes, it should be replaced_chars = len(''.join(map(lambda x: x[0], match))) instead