Remove invuca.icu table from the ICU data file. Cutting down the data size by...

Remove invuca.icu table from the ICU data file again. It was added to make Chrome not crash with a webkit patch for
bug 30437 (https://bugs.webkit.org/show_bug.cgi?id=30437)

However, the webkit patch was rolled back so that we don't need invuca.icu for now. Removing it will cut down the data size by 230k (before compression and in the Chrome executable on Linux/Mac and the ICU data dll on Linux).
This has to be added back when it becomes necessary again to fix the aforementioned webkit bug. 

See also http://crbug.com/28132 (a bug about ~ 4kB of space wasted by copyright statements) 

BUG=20406
TEST=NONE

http://src.chromium.org/viewvc/chrome?view=rev&revision=32640

[Evan Martin, 2009-11-20 18:50:01]

Do we expect the bug to be fixed again soon?  Is it worth cutting this out if we
plan to add it again?

[jungshik at Google, 2009-11-20 19:00:52]

On 2009/11/20 18:50:01, Evan Martin wrote:
> Do we expect the bug to be fixed again soon?  Is it worth cutting this out if
we
> plan to add it again?

I wrote a reply, but this tool crashed :-)

Anyway, the webkit bug will not be fixed for Chrome 4. So, for Chrome 4, we need
this. 

For Chrome 5, there's a way to fix the webkit bug without invuca.icu using a
newly proposed 'search collator' data entry in CLDR/ICU if the upstream can
accept a couple of #ifdef's in the webkit tree.

 Other ports of webkit that use ICU do not have the luxury of controlling the
ICU still need to use invuca.icu, but invuca.icu is available anyway. 

BTW, this CL does contain the 'assembly' sources of linux/mac icu data but
they're not shown here. Windows data dll will be landed separately.

[Evan Martin, 2009-11-20 19:07:14]

LGTM rubber stamp (I don't understand the build files, but the intent of change
sounds good)