Fix bug in scan merge optimization reported by Yao Li.

Support for a new type based on compyte in theano

Make tests try to run if they can.

Fix in local_dot22_to_dot22scalar

Fix the tutorial to make it work with python 32 and 64 bit.

Add tests and opt one more case.

new op wasn't added there.

Fix compilation crash reported in gh-1518 for gpujoin.

Be safer with the support code (in peculiar don't try to bake the C++ complex struct in the kernel code).

Also fix the call syntax for python <= 2.5

Fix the Cuda transfer ops and test them.  Also since the test will always fail unless initialized in a very specific way, mark it a potentially failing not to alarm users.

Add a dependency on libcompyte since we can't get access to the functions through the gpuarray module (dlopen is not global).

Update message to mention the need to clear the cache

BUGFIX: tutorial -> Derivatives in Theano -> Jacobian times a Vector -> L-operator.

Add tests for GpuFromCuda and CudaFromGpu (which I could not try).
Add tests for GpuAlloc inspired by the tensor tests.
While there adapt the makeTester function in the tensor tests for easier testing of other ops.