Support jumbo in chrome/browser, saving another 10% compile time

Of the code pieces left to add jumbo support to, the one which benefit
the most from jumbo is //chrome/browser:browser which goes from 90 to
10 CPU minutes, a reduction of more than a CPU hour. This is about 10%
of the total effort compiling chrome+content_shell+blink_tests.

In jumbo builds many cc files compile in the same translation unit
which reduces the time spent generating debug data and machine code
for headers, as well as the time to actually process the headers.

Bug: 746957
Change-Id: I856f3fca1f16fa2b0738678444c455c303985962
Reviewed-on: https://chromium-review.googlesource.com/899368Reviewed-by: Scott Violet <sky@chromium.org>
Commit-Queue: Daniel Bratell <bratell@opera.com>
Cr-Commit-Position: refs/heads/master@{#537284}

chrome/browser/BUILD.gn

@@ -5,6 +5,7 @@
 import("//build/config/chrome_build.gni")
 import("//build/config/crypto.gni")
 import("//build/config/features.gni")
+import("//build/config/jumbo.gni")
 import("//build/config/ui.gni")
 import("//build/split_static_library.gni")
 import("//chrome/common/features.gni")
@@ -70,7 +71,7 @@ if (enable_vr) {
 
 # Use a static library here because many test binaries depend on this but don't
 # require many files from it. This makes linking more efficient.
-split_static_library("browser") {
+jumbo_split_static_library("browser") {
   # Split into multiple static libraries on Windows builds. We have hit size
   # limits on Windows official builds and on goma builds when symbol_level = 2
   # is selected. Always splitting on Windows builds is simpler than trying to