Improved JavaScript performance for https://codereview.chromium.org//171773003

Improved performance for https://codereview.chromium.org//171773003

This version is 5x-8x faster than https://codereview.chromium.org//171773003
when timed on s.padLeft(6, '0') where s is 1, 3 or 7 characters.
5x for jsshell, 8x for v8.

[Lasse Reichstein Nielsen, 2014-02-20 06:45:08]

https://codereview.chromium.org/172543003/diff/30001/sdk/lib/_internal/lib/js...
File sdk/lib/_internal/lib/js_string.dart (right):

https://codereview.chromium.org/172543003/diff/30001/sdk/lib/_internal/lib/js...
sdk/lib/_internal/lib/js_string.dart:236: times = JS('JSUInt31', '# >>> 1',
times);
What if times is greater than 32 bits? (which would be silly, but  "" * 
10000000000000 is still valid).

[Lasse Reichstein Nielsen, 2014-02-20 06:51:33]

https://codereview.chromium.org/172543003/diff/30001/sdk/lib/_internal/lib/js...
File sdk/lib/_internal/lib/js_string.dart (right):

https://codereview.chromium.org/172543003/diff/30001/sdk/lib/_internal/lib/js...
sdk/lib/_internal/lib/js_string.dart:230: if (0 >= times) return '';  //
Unnecessary but hoists argument type check.
Not unnecessary since you use >>> below (and === 0 as guard). A negative number
would look positive after shifting.

https://codereview.chromium.org/172543003/diff/30001/sdk/lib/_internal/lib/js...
sdk/lib/_internal/lib/js_string.dart:236: times = JS('JSUInt31', '# >>> 1',
times);
In particular "0x100000000" (1p32) would be treated as zero, so  "a"*0x100000000
returns "". We should throw an out-of-memory error in this case, somehow.

[Lasse Reichstein Nielsen, 2014-02-20 11:50:31]

I merged this into my CL, added a test for too large numbers (throws an
OutOfMemoryError if times > 0xFFFFFFFF && this.isNotEmpty).

Feel free to optimize the test or reduce the range of valid "times". All JS
implementations will hit OOM long before that value.