Migrated several tests from blink to V8 repository.

test/webkit/regexp-non-bmp-expected.txt

Index: test/webkit/regexp-non-bmp-expected.txt
diff --git a/test/webkit/concat-while-having-a-bad-time-expected.txt b/test/webkit/regexp-non-bmp-expected.txt
similarity index 69%
copy from test/webkit/concat-while-having-a-bad-time-expected.txt
copy to test/webkit/regexp-non-bmp-expected.txt
index f949db98fd7b911d30b20281c82ee0f801405c77..28d78e2b003066a0f7478b4a1347b14d55620643 100644
--- a/test/webkit/concat-while-having-a-bad-time-expected.txt
+++ b/test/webkit/regexp-non-bmp-expected.txt
@@ -21,12 +21,20 @@
 # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 # SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-Tests the behavior of Array.prototype.concat while the array is having a bad time due to one of the elements we are concatenating.
+Tests that regular expressions treat non-BMP characters as two separate characters. From a Unicode correctness point of view this is wrong, but it is what other browsers do. And given that we store strings as UTF-16, it is also more efficient to implement. Also test some other cases related to UTF-8 and UTF-16.
 
 On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".
 
 
-PASS [42].concat() is [42]
+PASS /./.exec(surrogatePair).toString().length is 1
+PASS /\D/.exec(surrogatePair).toString().length is 1
+PASS /\S/.exec(surrogatePair).toString().length is 1
+PASS /\W/.exec(surrogatePair).toString().length is 1
+PASS /[^x]/.exec(surrogatePair).toString().length is 1
+
+PASS /.{1,2}/.exec("!!" + String.fromCharCode(0xA1)).toString().length is 2
+PASS /./.exec("") is null
+
 PASS successfullyParsed is true
 
 TEST COMPLETE