Include icudt38l_dat.s in icu

third_party/icu38/README.google

 A. This directory contains the source code of ICU 3.8 for C/C++ with CJK
   segmentation, plural formatting and other locale-data updates added:
 
   - It was obtained with the following:
 
     1. Download the clean ICU source tree (revision 22777) with the following :
 
        $ svn export --native-eol LF -r 22777 http://source.icu-project.org/repos/icu/icu/trunk icu38
 
     2. Apply the following 8 patches :
@@ skipping 30 matching lines @@
      
 
 B. We build ICU 3.8 with VC++ 8 statically linked to VC8 runtime library
   via the project files in the build directory.  The files under
   source\common\unicode and source\i18n\unicode were moved under public
   so that consumer don't accidently include internal header files.
 
 C. The data tables DLL (icudt38.dll) is converted from the archive file 
   (trunk/third_party/icu38/source/data/in/icudt38l.dat) built on Linux.
   Consumers normally just copy the converted file (icudt38.dll) by including
-  the build/icudt.vcproj file.
+  the build/icudt.vcproj file.  Additionally, icudt38l_dat.s is included as
+  prebuilt ICU data for the Mac build.
 
 D. The following procedure was taken to build a customized icudt38l.dat on Linux. 
 
    1. Download the ICU 3.8 source as explained in A.1 above.
 
      $ svn export --native-eol LF -r 22777 http://source.icu-project.org/repos/icu/icu/trunk icu38
 
    2. Apply the first two patches listed in A.2 above. 
    'data.patch.txt' contains the following data customization:
       
@@ skipping 92 matching lines @@
        source/test/intltest/tmsgfmt.h   
        public/common/unicode/utypes.h
        public/i18n/unicode/msgfmt.h
        public/i18n/unicode/plurfmt.h
        public/i18n/unicode/plurrule.h
 
   Besides, builds/icu.vcproj was modified to add plurrule.cpp and
   plurfmt.cpp to the list of files to compile. Note that this change
   is not a part of plural.patch because builds/icu.vcproj is 
   our own file.