Introduce a pydantic basemodel to manage input and output parameters, results

[mikibonacci]

class YamboParameters(BaseModel):
    BndsRnXp: t.Optional[t.List[int]] = Field(None, 
        description="Band range for the dynamical screening calculation")
    BndsRnXs: t.Optional[t.List[int]] = Field(None, 
        description="Band range for the static screening calculation")
    
class YamboInputs(BaseModel):
    runlevels: list
    parameters: YamboParameters

so that we can dump the model (defining rules to dump it following yambopy requirements) but also construct it from the parameters Dict, and manage easily all the parameters and so on.

This will improve the code for complex logic in convergences, mainly. But also help the users in define more easily the parameters, and see what he can possibly define.
Moreover, we can use the pydantic data validation