mojo/gpu: use SharedBuffer instead of base::SharedMemory with hacks

mojo/mojo_services.gypi

 {
   'targets': [
     {
       'target_name': 'mojo_gles2_bindings',
       'type': 'static_library',
       'sources': [
         'services/gles2/command_buffer.mojom',
         'services/gles2/command_buffer_type_conversions.cc',
         'services/gles2/command_buffer_type_conversions.h',
+        'services/gles2/mojo_buffer_backing.cc',
+        'services/gles2/mojo_buffer_backing.h',
       ],
       'variables': {
         'mojom_base_output_dir': 'mojo',
       },
       'includes': [ 'public/bindings/mojom_bindings_generator.gypi' ],
       'export_dependent_settings': [
         'mojo_bindings',
         'mojo_system',
       ],
       'dependencies': [
+        '../gpu/gpu.gyp:command_buffer_common',
         'mojo_bindings',
         'mojo_system',
       ],
     },
     {
       'target_name': 'mojo_gles2_service',
       'type': 'static_library',
       'dependencies': [
         '../base/base.gyp:base',
         '../gpu/gpu.gyp:command_buffer_service',
@@ skipping 65 matching lines @@
         }],
         ['OS=="android"', {
           'dependencies': [
             'mojo_jni_headers',
           ],
         }],
       ],
     },
   ],
 }