Fix empty list items being parsed as nested lists

[LuisFigueroaG]

Fixes #167

A list item with nothing after the marker (just the marker and a trailing space) took the next line as its content. That line then got parsed as a sub-list, so the output was nested and broken:

 1. 
 2. 

<ol>
<li>
<ol start="2">
2. 
</ol></li>
</ol>

- \n- \n did the same thing with <ul>, and an empty item in the middle of a list (- foo\n- \n- bar) turned bar into a sub-list of the empty item.

The item text in _list_item_re was (?:.+?), which needs at least one character, so for an empty item it went past the newline into the next item. I changed it to (?:.*?). For items that have text this matches the same as before. For empty items it now stops at the end of the line, so the output is:

<ol>
<li></li>
<li></li>
</ol>

That's also what GitHub gives for this input.

Tests:

• New case tm-cases/empty_list_items_issue167 covers the example from the issue, bullet lists, an empty item between other items, an empty numbered item and empty items in a sub-list. It fails without the fix.
• xss_smuggling_spans_in_image_attrs has a - line followed by - ![](...), and its expected HTML had the old nested-list output. I updated it to the two-item list. The <img> line itself is unchanged.

Test runs:

• make testone: 289 tests, OK (288 on master)
• make testredos: 7 tests, OK
• pycodestyle lib/markdown2.py gives the same count before and after (415)