Do not ignore SolidColorScrollbarLayer's scroll_layer_id ctor argument

cc/layers/solid_color_scrollbar_layer.cc

Index: cc/layers/solid_color_scrollbar_layer.cc
diff --git a/cc/layers/solid_color_scrollbar_layer.cc b/cc/layers/solid_color_scrollbar_layer.cc
index 2670fd3216541303f711bbbfcf22b8d98b3b3b29..2eb5484c855035419083a89ce65e147cff1a62f2 100644
--- a/cc/layers/solid_color_scrollbar_layer.cc
+++ b/cc/layers/solid_color_scrollbar_layer.cc
@@ -39,7 +39,7 @@ SolidColorScrollbarLayer::SolidColorScrollbarLayerInputs::
                                    int track_start,
                                    bool is_left_side_vertical_scrollbar,
                                    int scroll_layer_id)

[ajuma, 2017/03/04 00:04:50]

It looks like this gets called in non-test code with scroll_layer_id hardcoded
to 0: 
https://cs.chromium.org/chromium/src/cc/blink/web_scrollbar_layer_impl.cc?rcl...

The same happens for the other scrollbar layer subclasses (that do use the
scroll_layer_id argument) though. But it seems confusing, since 0 is not the
same as INVALID_ID (which is -1).

Would it make sense as part of this cleanup to change the callsites that are
passing in 0 when they mean INVALID_ID to actually use INVALID_ID?

[pdr., 2017/03/04 02:44:23]

We can actually do even better: these callsites are not needed at all. I have a
followup patch to remove them which cleans up a bunch of code:
https://codereview.chromium.org/2728253002

Mind if we land this patch as-is?

-    : scroll_layer_id(Layer::INVALID_ID),
+    : scroll_layer_id(scroll_layer_id),
       orientation(orientation),
       thumb_thickness(thumb_thickness),
       track_start(track_start),