We noticed the following whilst testing my project app.
If this is the node body <h1>HEADER<h1/>Lorem...
With 'correct faulty and chopped off html' enabled in the input filter, the output is <h1>HEADER</h1><h1></h1>Lorem...
With 'correct faulty and chopped off html' disabled in the input filter, the output is <h1>HEADER</h1><h1>Lorem... THE WHOLE NEXT SECTION BECOMES H1</h1>
The problem tested was the use of <h*/> instead of the correct </h*>. The HTML corrector must see the <h*> does not have an ending, then adds one before the start of the next tag. Maybe it should check if that tag is an incorrect ending?