Broadcast input of sparse_from_dense so ndim=2

Fix dtype of input value in test_diag

Since csc_matrix's default dtype is floatX, the numpy value needs
to have a compatible dtype.

Do not constant-fold Alloc if it is the output

make_node used to crash, now it works.
A test detects the problem.

This makes things easier for shape inference and gradient operations.
Also fix broadcasting pattern of the gradient.

Finished DOC/TEST/REVIEW sparse Diag Op

Fix test_diag_grad to work with verify_grad

Also fix an optimization bug in that case.

Fix perform of structureddot_grad with dense grad

Include "canonicalize" in tests for shape opt.

Fix comparison of "borrow" flag.

This was causing lots of warnings to print in the buildbot.
The error was introduced during a seemingly innocent PEP8 fix.

Update code for mean's dtype after code review

Miscellaneous documentation updates

Remove duplicated code.

This seems like the result of a bad merge.

Fixed crash with Composite support code and GPU

The problem was that when a Composite Op had a Composite subnode, having
the subnode return an empty support code would cause the parent node to
return '\n' as support code. As a result, GPU code generation would
crash due to non-empty support code.

Thanks to Fred for writing the associated test.

Make a machanism to don't have Alloc and GpuAlloc constant folded in som...