canvas: Specify size for RenderPass shared quad state and quad list size

CanvasResourceDispatcher creates a CompositorFrame with exactly one
quad, but because it uses the default RenderPass::Create() factory
function, it ends up allocating storage for 128 quads and 32 shared quad
states. A similar use case, VideoFrameSubmitter, was measured to take
around 0.2% of total Chrome CPU time in cases like Google Meet which
have a large number of VideoFrameSubmitters.  Although there are no
measurements for offscreen / low latency canvas, the same optimization
should help.

Bug: 1077299
Change-Id: I4862e8bda50edd35c73adf26c1d5b180de1cb2a1
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2181076
Commit-Queue: Sunny Sachanandani <sunnyps@chromium.org>
Commit-Queue: Khushal <khushalsagar@chromium.org>
Auto-Submit: Sunny Sachanandani <sunnyps@chromium.org>
Reviewed-by: Khushal <khushalsagar@chromium.org>
Cr-Commit-Position: refs/heads/master@{#765346}

third_party/blink/renderer/platform/graphics/canvas_resource_dispatcher.cc

@@ -230,7 +230,8 @@ bool CanvasResourceDispatcher::PrepareFrame(
   const gfx::Rect bounds(size_.Width(), size_.Height());
   constexpr int kRenderPassId = 1;
   constexpr bool is_clipped = false;
-  std::unique_ptr<viz::RenderPass> pass = viz::RenderPass::Create();
+  auto pass = viz::RenderPass::Create(/*shared_quad_state_list_size=*/1u,
+                                      /*quad_list_size=*/1u);
   pass->SetNew(kRenderPassId, bounds,
                gfx::Rect(damage_rect.x(), damage_rect.y(), damage_rect.width(),
                          damage_rect.height()),