[FEA]: Expose a fast modulo division algorithm for device code

[fbusato]

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

libcu++

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

Division and modulo with runtime values is a very common operation. The nature of the operations and GPU architectures makes the computation inefficient for device code.
On the other hand, division and modulo can be easily precomputed in the host. This allows efficient code for device computation.

Describe the solution you'd like

Expose a fast_int class to support all common integer operations and is optimized for modulo and division.

The following references should be considered:

• Classic Round-Up Variant of Fast Unsigned Division by Constants
• Faster remainders when the divisor is a constant
• CUB implementation
• cuCollection implementation

Describe alternatives you've considered

No response

Additional context

@sleeepyjack for visibility (cuCollection)