[FEA]: Specialize `cuda::std::expected` for `cudaError_t`

### 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.

We have a generic implementation of `cuda::std::expected` that works with all error types.

However, being so generic it is much harder on compile-times and also not as efficient as it could be for an error type such as `cudaError_t`

### Describe the solution you'd like

Rather than storing a boolean to check whether the value is valid we can just store an `cudaError_t` and signal `has_value` by comparing against `cudaSuccess`

That means that we avoid the additional storage for bool + cudaError_t and we can also simplify a lot of the member methods knowing that `cudaError_t` is a plain enumeration

### Describe alternatives you've considered

_No response_

### Additional context

_No response_