more work towards hybrid of GPU grid and sap broadphase, separating small, large (moving) and static objects

re-enable 'useNewBatchingKernel', it is slower but more robust

src/Bullet3OpenCL/RigidBody/b3Solver.cpp

@@ -17,7 +17,7 @@ subject to the following restrictions:
 #include "b3Solver.h"
 
 ///useNewBatchingKernel  is a rewritten kernel using just a single thread of the warp, for experiments
-bool useNewBatchingKernel = false;
+bool useNewBatchingKernel = true;
 bool convertConstraintOnCpu = false;
 
 #define B3_SOLVER_SETUP_KERNEL_PATH "src/Bullet3OpenCL/RigidBody/kernels/solverSetup.cl"