[FEA]: Refactor C++/C dispatch layers to avoid code duplication

[gevtushenko]

Is this a duplicate?

• [x] I confirmed there appear to be no duplicate issues for this request and that I agree to the Code of Conduct

Area

CUB

Is your feature request related to a problem? Please describe.

Today, C dispatch layer of parallel algorithms duplicates the logic in C++ dispatch layer. This is suboptimal, because we have to have extensive testing on both C and C++ ends to validate the logic, fix bugs and remember to introduce changes in multiple places.

Describe the solution you'd like

Research is needed to find ways to unify C++ and C dispatch layers.

Describe alternatives you've considered

No response

Additional context

No response