Due to size limit of shared memory, histogram is built on
the global memory for CV_16UC1 case.

The amount of memory needed for building histogram is:

    65536 * 4byte = 256KB

and shared memory limit is 48KB typically.

Added test cases for CV_16UC1 and various clip limits.
Added perf tests for CV_16UC1 on both CPU and CUDA code.

There was also a bug in CV_8UC1 case when redistributing
"residual" clipped pixels. Adding the test case where clip
limit is 5.0 exposes this bug.

* __shfl_up_sync with proper mask value for CUDA >= 9

* BlockScanInclusive for CUDA >= 9

* compatible_shfl_up for use in integral.hpp

* Use CLAHE in cudev

* Add tests for BlockScan

cuda_canny : multi stream safety (#11483)

* CUDA_ImgProc/Canny Asynchronous test

* cuda_canny : multi stream safety (1/3)

- Convert global variable canny::counter to class local variable

* cuda_canny : multi stream safety (2/3)

- Use texture objects rather than texture reference for cc >= 3.0,
  since texture reference must be declared as a static global variable
  which results in race condition when ran concurrently

* cuda_canny : multi stream safety (3/3)

- Refrain from using global variable in row_filter and column_filter
  (converts column_filter::c_kernel and row_filter::c_kernel to local
  variables)

* Fixes #11193

Found via `codespell`

Implement cv::cuda::calcHist with mask support (#8367)

* Implement cuda::calcHist with mask

* Fix documentation build warning

* Have their own step sizes for src and mask. Fix review comment.

use correct types for integral/sum outputs