Tweak HTML regexs

Tweak HTML regexs

For tags, need to do a non-consuming match rather than a consuming match
or else attributes may not be caught at the beginning of a tag.

Additionally, most browsers (except Opera) have trouble with tables set
to height="100%". In these cases, the calculated height will never be
proper since the table will always grow to the given height and will
hide any text below it.  We need to strip out these definitions (100%
height for tables doesn't make much sense anyway).