The original template based mat ptr for indexing is not implemented,
add the similar implementation as uchar type, but cast to
user-defined type from the uchar pointer.

Add constructors taking initializer_list for some of OpenCV data types (#9034)

* Add a constructor taking initializer_list for Matx

* Add a constructor taking initializer list for Mat and Mat_

* Add one more method to initialize Mat to the corresponding tutorial

* Add a note how to initialize Matx

* CV_CXX_11->CV_CXX11

Add support for std::array<T, N> (#8535)

* Add support for std::array<T, N>

* Add std::array<Mat, N> support

* Remove UMat constructor with std::array parameter

This test case uses a matrix with more dimensions than columns. Without
the fix in
https://github.com/opencv/opencv/pull/8448/commits/b45e784bebe318826bb48a41023db3fe1660d91e
this crashes with a segmentation fault, hangs or simply fails with wrong
values.

Currently, to select a submatrix of a N-dimensional matrix, it requires
two lines of code while only one line of code is required if using a 2D
array.

I added functionality to be able to select an N-dim submatrix using a
vector list instead of a Range pointer. This allows initializer lists to
be used for a one-line selection.

add template specialization Mat::push_back() for MatExpr paramters

extend push_back MatExpr to mat in unit test

cast to object instead of reference

test with multi-row MatExpr input

http://code.opencv.org/issues/2206,
http://code.opencv.org/issues/2924

fixed http://code.opencv.org/issues/3935
possibly fixed http://code.opencv.org/issues/3594

1. someMatrix.data -> someMatrix.prt()
2. someMatrix.data + someMatrix.step * lineIndex -> someMatrix.ptr( lineIndex )
3. (SomeType*) someMatrix.data -> someMatrix.ptr<SomeType>()
4. someMatrix.data -> !someMatrix.empty() ( or !someMatrix.data -> someMatrix.empty() ) in logical expressions

I propose forEach method for cv::Mat and cv::Mat_.
This is solution for the overhead of MatIterator_<_Tp>.

I runs a test that micro opecode runs all over the pixel of cv::Mat_<cv::Point3_<uint8_t>>.
And this implementation 40% faster than the simple pointer, 80% faster than iterator.
With OpenMP, 70% faster than simple pointer, 95% faster than iterator (Core i7 920).

Above all, code is more readable.

My test code is here.
    https://gist.github.com/kazuki-ma/8285876

Thanks.

Conflicts:

	modules/core/src/stat.cpp

Conflicts:
	modules/core/include/opencv2/core/cuda.hpp
	modules/cudacodec/src/thread.cpp
	modules/cudacodec/src/thread.hpp
	modules/superres/perf/perf_superres.cpp
	modules/superres/src/btv_l1_cuda.cpp
	modules/superres/src/optical_flow.cpp
	modules/videostab/src/global_motion.cpp
	modules/videostab/src/inpainting.cpp
	samples/cpp/stitching_detailed.cpp
	samples/cpp/videostab.cpp
	samples/gpu/stereo_multi.cpp

This and several following commits contain the "boring" changes required
to support the new Ptr. These are changes like:

* new T -> makePtr<T> or .reset(new T) or Ptr<T>(new T)
  (depending on the situation)
* p.empty() -> !p
* delete_obj -> DefaultDeleter::operator()

and similar changes that are numerous, but primitive.