Remove parameters metadata

[xjules]

Since all parameter models are pydantic there is no need to have an explicit metadata endpoint. Instead we directly call model_dump on the actual config.
Mostly relevant for:
src/ert/dark_storage/endpoints/experiments.py

[frode-aarstad]

I guess this means merging the ParameterData and ParameterConfig. Why were the ParameterMetadata there in the first place? Performance issues?

class ParameterMetadata(BaseModel):
    key: str
    transformation: str | None
    dimensionality: Literal[1, 2, 3] = 1
    userdata: dict[str, Any]


class ParameterConfig(BaseModel):
    type: str
    name: str
    forward_init: bool
    update: bool
...

[xjules]

I guess this means merging the ParameterData and ParameterConfig.

Yes, to some degree at least.
I think the dimensionality can be simply be a property of each parameter class; eg.

Update: It needs to be an attribute due to it should be present in the parameter configs in the storage.

All others can be removed as they are already in the config.

Why were the ParameterMetadata there in the first place? Performance issues?

Previously the parameter configs were not pydentic models. But it might be related to performance as well.

class ParameterMetadata(BaseModel):
    key: str
    transformation: str | None
    dimensionality: Literal[1, 2, 3] = 1
    userdata: dict[str, Any]


class ParameterConfig(BaseModel):
    type: str
    name: str
    forward_init: bool
    update: bool
...