Enable FP16 GPU target for DL Inference Engine backend.

* Add CPU default extensions loading in IE dnn backend

* Load cpu_extensions for the future Intel's Inference Engine

GDAL message:
- ERROR 1: The BMP file is probably corrupted or too large. Image width = 480

In line 104 `if ( full_neg_lst[i].cols >= box.width && full_neg_lst[i].rows >= box.height )` removed '=' as it causes divide By Zero Error in line 106 and 107 `box.x = rand() % ( full_neg_lst[i].cols - size_x );` when  full_neg_lst[i].cols = size_x or full_neg_lst[i].rows - size_y

Correct a mistake in {js,py}_table_of_contents_histograms.markdown (#11260)

* Update py_table_of_contents_histograms.markdown

"Learn to find and draw Contours" => "Learn the basics of histograms"

* Update js_table_of_contents_histograms.markdown

Corrected a seemlingly wronged line of copy in the python documention js_table_of_contents_histograms.markdown

"Learn to find and draw Contours" => "Learn the basics of histograms"

Also these samples require TBB, so they are not tested during default OpenCV builds

Regular OpenCV perf tests should be used instead

- properly detect Android SDK support of Ant/Gradle projects
- divide processing of And/Gradle based projects

- make non-copyable (aligns inner mutex semantic to std::mutex)
- getMemoryPool() returns reference instead of pointer (NULL is not expected here)