The HTML Processor understands HTML regardless of how it's written, but
many other functions are unable to do so. There are all sorts of syntax
peculiarities and semantics that would be helpful to eliminate using the
knowledge contained in the HTML Processor.

This patch introduces `WP_HTML_Processor::normalize( $html )` as a
method which takes a fragment of HTML as input and then returns a
serialized version of the input, "cleaning it up" by balancing all
tags, providing all missing optional tags, re-encoding all text,
removing all duplicate attributes, and double-quote-escaping all
attribute values.

Core-62036

Co-authored-by: Weston Ruter <westonruter@git.wordpress.org>

If code later in the processing pipeline adds unquoted attributes
and doesn't add the requisite space following that, then another
parser might find that the solidus is part of the attribute value
instead of serving as a self-closing flag.

Co-authored-by: Weston Ruter <westonruter@git.wordpress.org>

Co-authored-by: Weston Ruter <westonruter@git.wordpress.org>

Co-authored-by: Weston Ruter <westonruter@git.wordpress.org>

Co-authored-by: Weston Ruter <westonruter@git.wordpress.org>