htmLawed: Fix corruption to UTF8 encoded text

On some combinations of operating systems, PHP and libpcre versions, `\s`
will match the iso-8859-x non-breaking-space, 0xa0. This regular expression
will munge the UTF8 encoded version, 0xc2a0 to 0xc220, which is not a valid
UTF8 character.

When inserted into a UTF8 field in mysql, the text will be truncated at and
after the first invalid character.