joint (non-contact constraint) solver iterations is now working on GPU, but overall slower because of data copy.

Will move joint setup to GPU, and then some benefit should be visible.
Don't use 64 alignment, it causes data structures size mismatch between cpu and gpu

src/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.cpp

@@ -237,8 +237,8 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 			m_data->m_solver->solveContacts(m_data->m_narrowphase->getNumRigidBodies(),&hostBodies[0],&hostInertias[0],0,0,numJoints, joints);
 
 		}
-		//gpuBodies.copyFromHost(hostBodies);
-		printf("...\n");
+		gpuBodies.copyFromHost(hostBodies);
+		
 	}
 
 	if (numContacts)