Naming discrepancy between `env` in Processing and `environment` in Estimator

[mikeweltevrede]

Describe the feature you'd like

There is a discrepancy on passing environment variables in Processing and Estimator. The parameter is called env in Processing and environment in Estimator.

I would like these to be aligned. For backwards compatibility sake, this should probably be manifested through a environment_variables parameter, but any solution would work for me.

The problem

The problem is that we cannot have a unified interface to these entities using **kwargs to pass arguments without manually parsing a parameter ourselves.

Current situation van be something like this if we use env for both situations:

def data_processing(environment: Literal["dev", "preprod", "prod"], **kwargs):
    initialize_environment(environment)

    return Processing(**kwargs)

def model_training(environment: Literal["dev", "preprod", "prod"], **kwargs):
    initialize_environment(environment)

    env_vars = kwargs.pop("env")
    if env_vars:
        kwargs["environment"] = env_vars

    return Estimator(**kwargs)

data_processing(environment="dev", env={"MY_VAR": 42})
model_training(environment="dev", env={"MY_VAR": 67})

Or with a more compatible interface

def data_processing(environment: Literal["dev", "preprod", "prod"], **kwargs):
    initialize_environment(environment)

    env_vars = kwargs.pop("environment_variables")
    if env_vars:
        kwargs["env"] = env_vars

    return Processing(**kwargs)

def model_training(environment: Literal["dev", "preprod", "prod"], **kwargs):
    initialize_environment(environment)

    env_vars = kwargs.pop("environment_variables")
    if env_vars:
        kwargs["environment"] = env_vars

    return Estimator(**kwargs)

data_processing(environment="dev", environment_variables={"MY_VAR": 42})
model_training(environment="dev", environment_variables={"MY_VAR": 67})

Ideally, we would want it to look like this because both classes accept a environment_variables parameter:

def data_processing(environment: Literal["dev", "preprod", "prod"], **kwargs):
    initialize_environment(environment)

    return Processing(**kwargs)

def model_training(environment: Literal["dev", "preprod", "prod"], **kwargs):
    initialize_environment(environment)

    return Estimator(**kwargs)

data_processing(environment="dev", environment_variables={"MY_VAR": 42})
model_training(environment="dev", environment_variables={"MY_VAR": 67})