Make sure lexed node includes HTML all the way to matching closing tag

[mrpotes]

The lexer assumes an html block finishes at the next matching close tag, which is often not the case. Instead, the corresponding close tag should be found.

[chjj]

Duplicate of #236, and already solved with 8f705aa, however, it was never merged due to performance reasons. If I can optimize it, I'll merge it.

This PR looks like it might have the potential to be faster. However, compiling a regex every time is going to be incredibly slow. I'll try to optimize my code a bit more soon.

[mrpotes]

Yep, I didn't much like compiling a regex every time, but took comfort from the table, blockquote and list blocks all doing so. I couldn't think of a good way to avoid it though.

I wonder if you could do it without having to compile the regex every time by matching the open/close tag at the start and then backtracking the re.lastIndex by the length of the next open/close tag to allow you to recapture the tag name, without having to return to the start each time?

Something like:

/<\/?(tag)\/?>.*<(\/\1|\1(?:"[^"]*"|'[^']*'|[^'">])*?\/?)>/g