Use stack instead of malloc() for most calls to SkRasterPipeline::run().

Also split bench into run/compile variants to measure the effect:
 Before …f16_compile 1x  …f16_run 1.02x  …srgb_compile 1.56x  …srgb_run 1.61x
 After  …f16_run 1x  …f16_compile 1.01x  …srgb_compile 1.58x  …srgb_run 1.59x

CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD

Change-Id: I8e65fb2acdbb05ccc0b3894f16d7646603c3e74d
Reviewed-on: https://skia-review.googlesource.com/6621
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
diff --git a/bench/SkRasterPipelineBench.cpp b/bench/SkRasterPipelineBench.cpp
index 16dea8a..376efde 100644
--- a/bench/SkRasterPipelineBench.cpp
+++ b/bench/SkRasterPipelineBench.cpp
@@ -22,13 +22,18 @@
 //   - src = srcover(dst, src)
 //   - store src back as srgb/f16
 
-template <bool kF16>
+template <bool kF16, bool kCompiled>
 class SkRasterPipelineBench : public Benchmark {
 public:
     bool isSuitableFor(Backend backend) override { return backend == kNonRendering_Backend; }
     const char* onGetName() override {
-        return kF16 ? "SkRasterPipeline_f16"
-                    : "SkRasterPipeline_srgb";
+        switch ((int)kCompiled << 1 | (int)kF16) {
+            case 0: return "SkRasterPipeline_srgb_run";
+            case 1: return "SkRasterPipeline_f16_run";
+            case 2: return "SkRasterPipeline_srgb_compile";
+            case 3: return "SkRasterPipeline_f16_compile";
+        }
+        return "whoops";
     }
 
     void onDraw(int loops, SkCanvas*) override {
@@ -53,12 +58,20 @@
             p.append(SkRasterPipeline::to_srgb);
             p.append(SkRasterPipeline::store_8888, &dst_ctx);
         }
-        auto compiled = p.compile();
 
-        while (loops --> 0) {
-            compiled(0,0, N);
+        if (kCompiled) {
+            auto compiled = p.compile();
+            while (loops --> 0) {
+                compiled(0,0, N);
+            }
+        } else {
+            while (loops --> 0) {
+                p.run(0,0, N);
+            }
         }
     }
 };
-DEF_BENCH( return new SkRasterPipelineBench<true>; )
-DEF_BENCH( return new SkRasterPipelineBench<false>; )
+DEF_BENCH( return (new SkRasterPipelineBench< true,  true>); )
+DEF_BENCH( return (new SkRasterPipelineBench<false,  true>); )
+DEF_BENCH( return (new SkRasterPipelineBench< true, false>); )
+DEF_BENCH( return (new SkRasterPipelineBench<false, false>); )