[FEA]: Minimize group design for initial productization #12533

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.

Current design of thread groups has large API surface.

Describe the solution you'd like

This issue can be closed by a PR moving interfaces outside of minimal subset required by existing inference kernels into implementation-detail namespace.

Describe alternatives you've considered

No response

Additional context

No response