Use standardized accelerator-to-shortcut code for new backspace shortcut UI.

chrome/browser/ui/browser_window.h

Index: chrome/browser/ui/browser_window.h
diff --git a/chrome/browser/ui/browser_window.h b/chrome/browser/ui/browser_window.h
index a2eb2485ce0a048df67e24ac2654ca32e36faab1..814405f0d56d466d7bb5e1ace282dce6604c5f7d 100644
--- a/chrome/browser/ui/browser_window.h
+++ b/chrome/browser/ui/browser_window.h
@@ -164,7 +164,7 @@ class BrowserWindow : public ui::BaseWindow {
   // Shows a notice teaching the user the new shortcut for going back or forward
   // if the user has pressed the old shortcut more than once in three seconds
   // and the bubble has been shown less than five times.
-  virtual void MaybeShowNewBackShortcutBubble(bool forward) = 0;
+  virtual void MaybeShowNewBackShortcutBubble(int command_id) = 0;

[sky, 2016/11/29 00:14:54]

I like the boolean here (or use an enum) as it makes it clear there are only two
possibilities. Using a command_id means any int can be supplied, which isn't
what you want. How about leaving this (or change to enum)?

[Peter Kasting, 2016/11/29 00:58:38]

I thought about this some when making the change, but I thought it was kinda
better for the new_back_shortcut_bubble.cc code to be agnostic about the actual
commands.  It will actually work correctly with any command ID now, so if we
want to use it for some kind of more general case we can.

I dunno, it's not a huge deal either way, it just seemed slightly better?  I can
change it back if you feel strongly.

[sky, 2016/11/29 04:22:57]

It's certainly more general and it only works right now because the
implementation treats it as a boolean. By that I mean you have a conditional
checking for one command id and assuming everything is false. If someone called
this with the wrong thing they wouldn't know they were using it incorrectly.
I prefer the boolean.

 
   // Hides the new back shortcut bubble, if showing, by fading it out.
   virtual void HideNewBackShortcutBubble() = 0;