[BUGFIX] Properly match html tags with data-* attributes

The cropHTML method needs to account for html tags - therefore it must
match them. The regex used for matching tags misses html tags with
attributes that contain dashes. (\w matches "any 'word' character",
being "[…] any letter or digit or the underscore character […]") [1].

The current W3C html draft sets even less constraints on what makes an
attribute name [2], but including dashes looks like the most urgent
need.

[1] http://php.net/manual/en/regexp.reference.escape.php
[2] http://www.w3.org/TR/html51/syntax.html#attributes-0

Resolves: #63086
Releases: master, 6.2
Change-Id: I52832ca3e8d9660faa939e673af6cdc32f310b18
Reviewed-on: http://review.typo3.org/34371


Reviewed-by: Christian Kuhn <lolli@schwarzbu.ch>
Tested-by: Christian Kuhn <lolli@schwarzbu.ch>
Reviewed-by: Helmut Hummel <helmut.hummel@typo3.org>
Reviewed-by: Markus Klein <klein.t3@reelworx.at>
Tested-by: Markus Klein <klein.t3@reelworx.at>