[FEA]: consider to use `nv_exec_check_disable` for most of libcu++ algorithms / functors

### Is this a duplicate?

- [X] I confirmed there appear to be no [duplicate issues](https://github.com/NVIDIA/cccl/issues) for this request and that I agree to the [Code of Conduct](CODE_OF_CONDUCT.md)

### Area

libcu++

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

Currently we mark everything as `__host__ __device__` which would disallow `__device__` or `__host__` only functions even if they are perfectly fine and only ever used in the right execution context.

We should add `nv_exec_check_disable` to most of those features that take a function object so that we do not bother users too much with it

### Describe the solution you'd like

Add the appropriate macro consistently in libcu++

### Describe alternatives you've considered

_No response_

### Additional context

_No response_