Remove the assert in PyGpuArray_DEV_DATA since it might cost a bit too much to verify that condition everytime.

Should fix bugs in current uses and makes porting easier too.

Update the subtensor test to have varying values in order to really check that it writes the right thing at the right spot.

Use the libgpuarray APIs to manage GPU code compilation, execution, etc.

Use the testcode facility to test code examples directly

All tests pass!