Fix an apparent typo in a unittest that caused it to not actually check the new window list it fetched.

This would only have an effect when window capture failed, and apparently that was never happening.

BUG=none
R=sergeyu@chromium.org

Review URL: https://webrtc-codereview.appspot.com/52009004

Cr-Commit-Position: refs/heads/master@{#9348}
diff --git a/webrtc/modules/desktop_capture/window_capturer_unittest.cc b/webrtc/modules/desktop_capture/window_capturer_unittest.cc
index bb4ddf0..8b38bcd 100644
--- a/webrtc/modules/desktop_capture/window_capturer_unittest.cc
+++ b/webrtc/modules/desktop_capture/window_capturer_unittest.cc
@@ -75,8 +75,8 @@
     if (!frame_.get()) {
       WindowCapturer::WindowList new_list;
       EXPECT_TRUE(capturer_->GetWindowList(&new_list));
-      for (WindowCapturer::WindowList::iterator new_list_it = windows.begin();
-           new_list_it != windows.end(); ++new_list_it) {
+      for (WindowCapturer::WindowList::iterator new_list_it = new_list.begin();
+           new_list_it != new_list.end(); ++new_list_it) {
         EXPECT_FALSE(it->id == new_list_it->id);
       }
       continue;