[FEA]: Implement cuda.parallel.{sum,min,max) algorithms

[gevtushenko]

Is this a duplicate?

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

Area

Thrust

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

Today, we have cuda.parallel.reduce algorithm which always accepts user-defined operators. This works, but deprives Python users from optimizations we have in special cases for {sum,min,max} operators, so we should expose those.

Describe the solution you'd like

Asynchronous `cuda.parallel.{sum,min,max} algorithms.

Describe alternatives you've considered

No response

Additional context

No response

[gevtushenko]

#3574 should be addressed first