cuda - update npp calls to use the new NppStreamContext API if available

[cudawarped]

Requires changes in opencv/opencv#26245

Update calls to Nvidia's NPP library to use the newer NppStreamContext API introduced in CUDA 10.1 (release notes)

Although this was introduced in CUDA 10.1 at that point not all of the functions either had a _Ctx version which took a NppStreamContext or an implementation present in the NPP library. Therefore in this PR the USE_NPP_STREAM_CTX flag which selects the new API is only enabled for versions of NPP >= 12205 (the version included with CUDA toolkit 12.4) which I have verified as working. Currently the only function called inside OpenCV which does not work with the new API is nppiEvenLevelsHost_32s_Ctx which doesn't have an implementation inside nppist.lib or libnppist.so as of CUDA 12.6.

Tested locally on Ubuntu 20.04 with both CUDA toolkits 12.3 and 12.4.

Adds changes discussed in #3338 to all Npp functions called by OpenCV.
Fixes #3798

@tomoaki0705 I have re-introduced streaming in https://github.com/tomoaki0705/opencv/blob/f4e5d777e856f751c4318f0c633dcce37cfa66f2/modules/cudaarithm/src/reductions.cpp#L161 which you disabled in opencv/opencv#11064 because it looks like you fixed it with opencv/opencv#11550 is this correct?

Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

• I agree to contribute to the project under Apache 2 License.
• To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
• The PR is proposed to the proper branch
• There is a reference to the original bug report and related work
• There is accuracy test, performance test and test data in opencv_extra repository, if applicable
  Patch to opencv_extra has the same branch name.
• The feature is well documented and sample code can be built with the project CMake

[cudawarped]

@asmorkalov Would it be possible to run this on the CI?

[cudawarped]

@asmorkalov I've just noticed you've merged opencv/opencv#2624.

I should have mentioned in that PR that both that and this PR are dependant on each other. Because NPP streams were previously global entities, only merging opencv/opencv#26245 breaks streaming for all OpenCV functions which use NPP under the hood when OpenCV is compiled against CUDA >=12.4.