Inductor cpp wrapper: support more dtypes of input (#88666)
Previously only float32 is supported as input types for the cpp wrapper.
This PR extends the cpp wrapper to support the built-in types: float32, float64, int64, int32, int16, int8, uint8, bool.
Bfloat16 and Float16 will be covered later.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/88666
Approved by: https://github.com/jgong5, https://github.com/jansel
