Yes, it's as ludicrous as it sounds, but it's still true. Bizarrely,
the previous commit makes CLAHE run about 10% slower on Android, even
though it doesn't even touch any CLAHE code. Splitting it off fixes that,
although the reason it does is a mystery for the ages.

It's cleaner when it's in its own file, anyway. ;=]

This should allow many algorithms to take advantage of more parallelization
technologies.

The orientation of convexHull's result is currently the opposite of what the
documentation would suggest:

>>> import cv2, numpy as np
>>> points = np.array([[0,0],[0,1],[1,0]], dtype=np.int32)
>>> cv2.convexHull(points, clockwise=False)
array([[[1, 0]],
       [[0, 1]],
       [[0, 0]]], dtype=int32)
>>> cv2.convexHull(points, clockwise=True)
array([[[0, 0]],
       [[0, 1]],
       [[1, 0]]], dtype=int32)

Changing the function itself is probably not a good idea at this point, so
this fixes the documentation by flipping the coordinate system.

I also removed the mention of the origin, since it's irrelevant for this
function.

Also cv::, cv::gpu:: and cv::ocl:: namespace prefixes can be safely omitted
inside CV_ENUM and CV_FLAGS

bug #2709

Remove unused var "total"
Declare vars as locally as possible

- add syntetic images generation function to ts
- use generated syntetic images

issue #2678

On some platforms sin and cos are calculated in single precision resulting in
diversity of results.

Old implementation did

    lut[i] = 255 * (count(Y <= i)) / (width * height)

which actually shifts uniform histograms.
From now histogram is equalized as

    C = count(Y == min(Y))
    lut[i] = 255 * (count(Y <= i) - C) / (width * height - C)