es53 css sanitizer barfs on <global-name>

<style>p { animation-name: foo; }</style>
throws a SomethingWidgy error in es53.

That's very unfriendly, because the guest-code developer ends up
getting a "Server Error" message with no diagnostic details,
and it's not at all obvious what to do next to fix it.

Since css rules are loaded dynamically at run-time, we can't really
know at compile-time what all the rule atoms are, so encountering an
unrecognized atom doesn't really merit a SomethingWidgy error.

Instead, this CL just fails the match, which makes CSS validator
emit a warning about a bad value, drop the value, and proceed.
And the guest-code developer can find the warning easily.

[MikeSamuel, 2013-08-15 00:11:46]

lgtm

https://codereview.appspot.com/12837046/diff/1/src/com/google/caja/plugin/Css...
File src/com/google/caja/plugin/CssValidator.java (right):

https://codereview.appspot.com/12837046/diff/1/src/com/google/caja/plugin/Css...
src/com/google/caja/plugin/CssValidator.java:1347: // Unrecognized atom: fail
the match.
I guess the checks you handled to CssPropertyPatterns will still help us suss
out problems like this statically in the main branch and this degrades more
gracefully

https://codereview.appspot.com/12837046/diff/1/tests/com/google/caja/plugin/C...
File tests/com/google/caja/plugin/CssValidatorTest.java (right):

https://codereview.appspot.com/12837046/diff/1/tests/com/google/caja/plugin/C...
tests/com/google/caja/plugin/CssValidatorTest.java:50: "WARNING: css property
animation-name has bad value: ==>foo<==");
Ok.  Maybe a comment that foo could be ok, but foo__ should not be.

[felix8a, 2013-08-19 22:30:21]

@r5564