[CUCO][FEA]: Migrate host and device `erase` APIs for fixed_capacity_map

### 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

CUDA Experimental (cudax)

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

This is part of #7463.

I'd like to migrate the host and device `erase` APIs from `cuco::static_map` to `cudax::cuco::fixed_capacity_map`.

### Describe the solution you'd like

Migrate the APIs and add corresponding tests and benchmarks.

Compare the host and device erase APIs in cuco::static_map with cudax::cuco::fixed_capacity_map. Add the migrated APIs, corresponding tests, and benchmarks; completion means both host and device behavior are covered and benchmark results are available.

Written by the indexing model from the issue text.