use cv::AutoBuffer::data() to get data pointer

This is to achieve parity with the CPU equivalent.

Based on commit feb74b12 from 2.4.9 branch.

changed tests for tvl1 optflow
correction of a bug preventing compilation with cuda (fmin changed to fminf)

Now multiple filters can be used in the same context without stepping on each other.

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