I caught places where the warning about using grad w.r.t. a one-element
list was raised when running the test-suite.
This commit removes these warnings (ensuring changing the default value of
keep_wrt_type to True will not affect tests).

Note that I changed the behavior of 'test_dot_w_self'. I asked the
author (James) before doing so: it turns out it was not behaving as
intended before.

__hash__ should take only 1 argument.

Added new option 'keep_wrt_type' to tensor.grad

This allows one to decide which behavior is desired (future new behavior
or old one) and get rid of the warning at the same time.

Print the name of the optimizer when we have it.

We disable amdlibm for nvcc. This was previously done in half.

Fix

This is a strange case as we restect the 256 bytes limits!

fix test in float32.

added arccos

Deprecated cache entries now trigger a warning

Most people should have updated their local cache by now. Making it a
warning will ensure we notice if for some reason a bug generates new
invalid cache entries.

Misc style fixes in op.py

Functions for clean interactive debugger code

interactive debugger

Deprecated tensor.shared (renamed tensor._shared)

STY: various blas/test_blas fixes

Most of the time people who use tensor.shared do it because they mix it up
with theano.shared. By renaming it with a leading underscore, this confusion
should disappear.

By the way this commit also changes a few calls in Theano code from
tensor.shared to theano.shared, when I thought calling tensor.shared was not
intentional.

Fix missing pydot (tests/error message)