Gamepad: honor page visibility

Gamepad: honor page visibility

This CL attempts to make gamepad events work perfectly in relation to page
visibility state in all cases.

Now we stop listening on gamepad events when page is hidden. It did not work
because DeviceSensorEventController did not tell NavigatorGamepad to
unregister if there were no gamepad event listeners. But since gamepad data can
be polled without a listener this was wrong. When we become visible again we
tell the page about all changes. Note that handling all this at the browser side
would be impractical because it would need to track which changes were observed
by which renderer (think about multiple tabs listening and the user switches 
between them).

NavigatorGamepad::didRemoveEventListener now considers that there might be
multiple listeners.

Finally as a refactor now we use DeviceSensorEventController::dispatchDeviceEvent
in NavigatorGamepad instead of duplicating some of it's logic.

BUG=344556

[kbalazs, 2014-06-02 17:57:29]

On 2014/06/02 17:52:55, kbalazs wrote:

<off>
I hate that rietveld truncates my description
</off>

[bajones, 2014-06-02 18:52:25]

On 2014/06/02 17:57:29, kbalazs wrote:
> On 2014/06/02 17:52:55, kbalazs wrote:
> 
> <off>
> I hate that rietveld truncates my description
> </off>

Non-owner LGTM. Thanks for the fix/cleanup.

[kbalazs, 2014-06-02 20:48:54]

On 2014/06/02 18:52:25, bajones wrote:
> On 2014/06/02 17:57:29, kbalazs wrote:
> > On 2014/06/02 17:52:55, kbalazs wrote:
> > 
> > <off>
> > I hate that rietveld truncates my description
> > </off>
> 
> Non-owner LGTM. Thanks for the fix/cleanup.

Btw it depends on https://codereview.chromium.org/304403002, the tests would
fail without it.

[kbalazs, 2014-06-04 20:39:03]

On 2014/06/02 20:48:54, kbalazs wrote:
> On 2014/06/02 18:52:25, bajones wrote:
> > On 2014/06/02 17:57:29, kbalazs wrote:
> > > On 2014/06/02 17:52:55, kbalazs wrote:
> > > 
> > > <off>
> > > I hate that rietveld truncates my description
> > > </off>
> > 
> > Non-owner LGTM. Thanks for the fix/cleanup.
> 
> Btw it depends on https://codereview.chromium.org/304403002, the tests would
> fail without it.

I might want to add some more craft to this, so wip yet. Feedback could be
useful though.

[kbalazs, 2014-06-05 22:46:48]

This is now ready together with https://codereview.chromium.org/304403002/
(which is needed for the tests).

[kbalazs, 2014-06-12 23:59:22]

On 2014/06/05 22:46:48, kbalazs wrote:
> This is now ready together with https://codereview.chromium.org/304403002/
> (which is needed for the tests).

This is no outdated and will need to be rebased on top of
https://codereview.chromium.org/336693004/ (assuming it will land in some form).

[kbalazs, 2014-06-16 22:57:25]

On 2014/06/12 23:59:22, kbalazs wrote:
> On 2014/06/05 22:46:48, kbalazs wrote:
> > This is now ready together with https://codereview.chromium.org/304403002/
> > (which is needed for the tests).
> 
> This is no outdated and will need to be rebased on top of
> https://codereview.chromium.org/336693004/ (assuming it will land in some
form).

Let's close this, rework will come in other cl.