Reverse accidental enabling of 'use strict' in closure compiler

tools/grit/grit_rule.gni

Index: tools/grit/grit_rule.gni
diff --git a/tools/grit/grit_rule.gni b/tools/grit/grit_rule.gni
index 3db7a6c6b210ceb8826faaef6ae3535efb26e601..ca524b6e287df5fb9b18add95f39fa8bb2181565 100644
--- a/tools/grit/grit_rule.gni
+++ b/tools/grit/grit_rule.gni
@@ -402,7 +402,10 @@ template("grit") {
         "--js-minifier",
         js_minifier_command,
       ]
-      inputs += [ _js_minifier ]
+      inputs += [
+        _js_minifier,
+        "//third_party/closure_compiler/compiler/compiler.jar",

[flackr, 2017/06/05 17:39:46]

I'm not sure I follow this change, is this invalidating compiled resources on
the compiler changing? Does the compiler change when you change the arguments
set in closure_arguments.gni?

[wychen, 2017/06/05 17:46:25]

The dependency was underspecified, so that when closure rolled in, the CQ and
waterfall didn't reliably repack the resources, and made bisecting difficult for
crbug.com/728655.

[flackr, 2017/06/05 17:54:54]

Ah I see, this makes sense to do as a separate commit. You may also want to
depend on the arguments file right?

[wychen, 2017/06/05 18:25:56]

I'll create a separate CL.
What do you mean by argument file? It looks like this js minifier use command
line arguments, and I think it'll get rebuilt if the arguments change.

There are several python scripts that's underspecified as well. Not sure if it
makes sense to add them all.

[flackr, 2017/06/05 18:46:40]

Ah okay, as long as argument changes in closure_args.gni trigger a rebuild then
it should be fine. I'm not sure, about adding other dependencies, I would stick
to just the ones which are known to be problematic.

[wychen, 2017/06/05 19:05:43]

Split out here:
https://codereview.chromium.org/2924653002/

+      ]
     }
 
     # Must be after the args are computed since they are re-used.