Correct gyp toolsets for all combinations of want_separate_host_toolset and use_system_icu.

Correct gyp toolsets for all combinations of want_separate_host_toolset and use_system_icu.

This CL ensures following logics:
if want_separate_host_toolset == 0 and use_system_icu == 0
  don't use system icu
  build icu for target toolset

elif want_separate_host_toolset == 0 and use_system_icu == 1
  use system icu for target toolset
  don't build icu

elif want_separate_host_toolset == 1 and use_system_icu == 0
  don't use system icu
  build icu for host and target

elif want_separate_host_toolset == 1 and use_system_icu == 1
  use system icu for target toolset
  build icu for host toolset

All complexity is due to the condition: want_separate_host_toolset == 1 and use_system_icu == 1.
Although use_system_icu == 1, this condition requires to build icu for host toolset.

BUG=
R=jochen@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=244134

[jochen (gone - plz use gerrit), 2014-01-07 10:16:02]

generally looks good

can you verify that this builds on the various configurations?

android_aosp has use_system_icu and want_separate_host_toolset enabled

[dshwang, 2014-01-09 11:50:20]

On 2014/01/07 10:16:02, jochen wrote:
> generally looks good
> 
> can you verify that this builds on the various configurations?
> 
> android_aosp has use_system_icu and want_separate_host_toolset enabled

Hi, thx for review.

Unfortunately, trybot does not support thirdparty,
so I built android_aosp in my ubuntu pc. I works well.

In detail, I built it as follows:
export GYP_GENERATOR_FLAGS="output_dir=out_android"
. build/android/envsetup.sh
android_gyp -Duse_system_icu=1 -Dwant_separate_host_toolset=1
ninja -C out_android/Release -j 10 content_shell_apk android_webview_apk

[jochen (gone - plz use gerrit), 2014-01-09 11:52:42]

lgtm

[dshwang, 2014-01-09 16:10:46]

On 2014/01/09 11:52:42, jochen wrote:
> lgtm

Thank you for lgtm. I guess there is not a commit bot. Could I ask how to land
this CL?

[inactive_dshwang_plz_cc_intel, 2014-01-10 09:23:44]

Committed patchset #1 manually as r00f025a (presubmit successful).

[dshwang, 2014-01-10 09:28:46]

On 2014/01/10 09:23:44, dongseong.hwang wrote:
> Committed patchset #1 manually as r00f025a (presubmit successful).

I try to land it by 'git cl push', but it failed (falsely committed) although
the CL looks committed.
Could you help me land it?

[jochen (gone - plz use gerrit), 2014-01-10 09:36:19]

Committed patchset #1 manually as r244134 (presubmit successful).

[jochen (gone - plz use gerrit), 2014-01-10 09:36:49]

On 2014/01/10 09:36:19, jochen wrote:
> Committed patchset #1 manually as r244134 (presubmit successful).

I committed it, no problem

(we still use svn for most things, so git cl push will never work)

[jochen (gone - plz use gerrit), 2014-01-10 09:37:10]

can you post a CL to update the chromium deps?

[dshwang, 2014-01-10 10:19:24]

On 2014/01/10 09:37:10, jochen wrote:
> can you post a CL to update the chromium deps?

Thank you! yes, I'll do it.

[dshwang, 2014-01-10 10:23:08]

On 2014/01/10 10:19:24, dshwang wrote:
> On 2014/01/10 09:37:10, jochen wrote:
> > can you post a CL to update the chromium deps?
> 
> Thank you! yes, I'll do it.

I see the your commit. If the commit has something like 'authored by dongseong',
it would be better.
I'm fine, no problem.

[jochen (gone - plz use gerrit), 2014-01-10 10:38:51]

On 2014/01/10 10:23:08, dshwang wrote:
> On 2014/01/10 10:19:24, dshwang wrote:
> > On 2014/01/10 09:37:10, jochen wrote:
> > > can you post a CL to update the chromium deps?
> > 
> > Thank you! yes, I'll do it.
> 
> I see the your commit. If the commit has something like 'authored by
dongseong',
> it would be better.
> I'm fine, no problem.

yeah, I tried to add that, but somehow it failed :-/

sorry about that