Improve handling of <video>, <figure> and <br>

Improve handling of <video>, <figure> and <br>

Omit <video> and <figure> from the document Boilerpipe sees as they have
text that can confuse processing. Instead treat these (and <br>s) in a
similar manner to data tables and <img> tags.

This shows a very minor decline in eval but it's a false negative as a
<figcaption> is omitted from the text only response, but is included
later when reconstructing the page.

BUG=376107, 376102, 378385
R=cjhopman@chromium.org, nyquist@chromium.org

Committed: b6100c8

[cjhopman, 2014-06-06 23:06:23]

lgtm

[nyquist, 2014-06-06 23:40:36]

lgtm

[Yaron, 2014-06-09 16:34:21]

Committed patchset #2 manually as rb6100c8 (presubmit successful).