Don't make |field_| first responder in SetSelectedRange() unless |model_| has_focus already.

Don't make |field_| first responder in SetSelectedRange() unless |model_| has_focus already.

http://crbug.com/11920
TEST=Browser to www.google.com, focus is in search field, bring up new tab, click back to first tab, focus should be in search field.
TEST=Select messages in gmail, focus should NOT go to autocomplete field (omnibox, url bar).

[Scott Hess - ex-Googler, 2009-05-22 23:16:48]

As noted in the bug, I had this part of the fix but was trying to fix a harder
part of the problem.  There was a #chromium complaint about a horrible variant
of the problem I didn't know about, so I figure it's time to put this up and
move the rest of the issue into new bugs.

[John Grabowski, 2009-05-22 23:32:18]

LGTM

In the bigger picture, it would be nice if all the focus rules were laid out
nicely and together in both doc and code to avoid odd bugs like this.  Can you
spend some time thinking if that is feasable?

http://codereview.chromium.org/113751/diff/1003/1005
File chrome/browser/autocomplete/autocomplete_edit_view_mac.mm (right):

http://codereview.chromium.org/113751/diff/1003/1005#newcode290
Line 290: if (model_->has_focus()) {
Logic of comment is opposite of conditional.
Reading it at face value makes me think you are missing a !.
Perhaps you can continue with an 
// If it does, switch focus to the current window.

[Scott Hess - ex-Googler, 2009-05-23 00:10:26]

I'll try to write my understanding up somewhere once I have it.  Keeping
autocomplete selection right across tab-switching will probably require my brain
to contain that.

http://codereview.chromium.org/113751/diff/1003/1005
File chrome/browser/autocomplete/autocomplete_edit_view_mac.mm (right):

http://codereview.chromium.org/113751/diff/1003/1005#newcode290
Line 290: if (model_->has_focus()) {
On 2009/05/22 23:32:18, John Grabowski wrote:
> Logic of comment is opposite of conditional.
> Reading it at face value makes me think you are missing a !.
> Perhaps you can continue with an 
> // If it does, switch focus to the current window.

I dropped the comment entirely, the header-file comment for the function made
more sense.

Updated the TODO before FocusLocation(), as I think things are getting tighter
on that front, now.