Issue 5008 fixed

Add bool dtype in scalar and tensor.

Fixed typo in config doc

make var() return float16 when input is float16.

Use "PyArray_Transpose" in gemm implementation

Add a test to test_blas.py:TestBlasStrides to check
is gemm works well with non-contiguous matrices
(A,B,C are all passed as non-contiguous).

All tests passed ! Recall: these are the tests executed:

theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/nnet/tests/test_abstract_conv.py:TestAbstractConvNoOptim
theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/nnet/tests/test_abstract_conv.py:TestBilinearUpsampling
theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/nnet/tests/test_abstract_conv.py:TestCorrConv2d
theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/nnet/tests/test_abstract_conv.py:TestCorrConv3d
theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/nnet/tests/test_abstract_conv.py:TestCpuConv2d
theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/nnet/tests/test_abstract_conv.py:TestCpuConv3d
theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/tests/test_blas_c.py
theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/tests/test_blas.py
theano-cache purge && THEANO_FLAGS=blas.ldflags= nosetests theano/tensor/tests/test_blas_scipy.py
theano-cache purge && THEANO_FLAGS=optdb.max_use_ratio=10,blas.ldflags= nosetests theano/tensor/nnet/tests/test_corr3d.py
theano-cache purge && THEANO_FLAGS=optdb.max_use_ratio=7,blas.ldflags= nosetests theano/tensor/nnet/tests/test_corr.py

Gemm implementation rewritten so that a new data memory
is allocated when we can not reuse C.

C can not be reused either when BETA != 0,
or when BETA == 0 but C is not contiguous
(LDC != M).

All tests passed, + special test suggested by lamblin.

Insert GpuSolve only when CULA is available

Doc update about cudnn and cudnn RNN.

Code-comment removed from test_corr.
Code factorized in alt_gemm_template.c

All tests passed.

Some commented codes about empty ldflags have been removed
and replaced with normal comments.

Some changes are made in alt_gemm_template.c relative to
last recommendations.

Fix a problem where the samples buffer would not be filled.

Sum with a complex input and scalar output doesn't work because of the
ambiguity of casting complex to scalar.  The reverse doesn't work
because of the ambiguity of the casting in the gradient.  Therefore
disallow both.

* Removed other checkings of empty blas.ldflags in blas.py
and test_blas.py.

* Correction in alt_gemm_template.c to prevent some test
failures in test_blas.py (occuring after above modification):
in some tests, a zero-content matrix is passed as C
(with M*N == 0). Now if we encounter this case, we just
skip it (gemm not calculated, C not modified).

* Correction in blas_headers.py to prevent some old string-
formating errors when strings are C code containing "%" symbols.

Now test_blas.py runs perfectly. I have to re-run all other
tests tonight.

New optimization: DimShuffle(Alloc(...)) -> Alloc(..., use dims of 1 for new dimensions) #5024

Log sum exp optimization for numerical stability

Manual revert of 29e67107