DM: make GPU tasks multithreaded again. Big refactor.

dm/DMTileGridTask.h

 #ifndef DMTileGridTask_DEFINED
 #define DMTileGridTask_DEFINED
 
 #include "DMTask.h"
 #include "SkBitmap.h"
 #include "SkString.h"
 #include "SkTemplates.h"
 #include "gm.h"
 
 // Records a GM through an SkPicture, draws it in tiles, and compares against the reference bitmap.
 
 namespace DM {
 
-class TileGridTask : public Task {
+class TileGridTask : public CpuTask {
 
 public:
     TileGridTask(const Task& parent,  // TileGridTask must be a child task.  Pass its parent here.
                  skiagm::GM*,         // GM to run through a picture.  Takes ownership.
                  SkBitmap reference,  // Bitmap to compare picture replay results to.
                  SkISize tileSize);   // Tile size to use.
 
     virtual void draw() SK_OVERRIDE;
-    virtual bool usesGpu() const SK_OVERRIDE { return false; }
     virtual bool shouldSkip() const SK_OVERRIDE;
     virtual SkString name() const SK_OVERRIDE { return fName; }
 
 private:
     const SkString fName;
     SkAutoTDelete<skiagm::GM> fGM;
     const SkBitmap fReference;
     const SkISize fTileSize;
 };
 
 }  // namespace DM
 
 #endif  // DMReplayTask_DEFINED