HTML examples seen in the wild that cause parse errors
in HTMLParser include:

<a width="100%"cellspacing=0>
  -- note lack of space between val and next attr name

<a foo=>
  -- trailing attribute has no value after =

<a href=javascript:popup('/popup/html.html')>
  -- javascript fragment with embedded quotes

My patch contains improvements to the 'attrfind' and
'locatestarttagend' regexps that allow these examples
to parse.

The existing test_htmlparser.py unit test continues to
pass, except for the one test case where it considers
<a foo=> to be an error.

I commented out that case and added new test cases to
cover the examples above.